Campaigns

Weekly #271 Digital Cinderellas: When the clock strikes midnight online

10 – 17 July 2026

HIGHLIGHT OF THE WEEK

Digital Cinderellas: When the clock strikes midnight online

For centuries, midnight has been the moment when magic ends. Cinderella’s carriage turns back into a pumpkin, the ball is over, and the spell is broken. The same symbolism is now present in the digital age, as the UK suggest that teenagers should also leave the online ball when the clock strikes twelve.

The UK has proposed a default social media curfew for 16- and 17-year-olds between midnight and 6 a.m. Platforms would pause personalised feeds and other engagement features by default, although teenagers could choose to opt out. The shift reflects an attempt to nudge healthier online habits without imposing an outright prohibition.

The midnight curfew may not have emerged out of thin air. It might also be inspired by a recent government-backed study that tested three approaches to limiting teenagers’ social media use: a 15-minute daily limit per app, an overnight curfew from 9 p.m. to 7 a.m., and the complete removal of selected social media apps. Participants reported better sleep, improved concentration, and higher overall well-being, providing policymakers with evidence that limiting digital engagement can produce tangible well-being benefits.

There are good reasons why such an approach appeals to policymakers. Social media platforms are designed to maximise engagement, and late-night scrolling is widely associated with poorer sleep and reduced well-being. Default bedtime settings could help interrupt the endless cycle of ‘just one more video’ without preventing young people from accessing online services altogether.

At the same time, the proposal raises important questions. Because the curfew is voluntary, many teenagers may simply disable it, limiting its effectiveness. Others question whether a single rule makes sense for all 16- and 17-year-olds, many of whom study late, work evening shifts, or communicate across time zones. Critics have also pointed to the contradiction between treating older teenagers as increasingly autonomous citizens (Labour has proposed lowering the voting age to 16) while simultaneously introducing special restrictions on their online behaviour.

Book, Publication, Comics, Baby, Person, Art, Face, Head

The UK’s softer approach contrasts with Australia’s far more forceful strategy. Canberra has famously been the first to introduce a ban on social media for under-16s, but early testing has exposed significant practical obstacles. A follow-up study by researchers involved in the government’s age-assurance trials found that platforms rarely requested proof of age from users who simply declared themselves to be 16 or older. None of the 50 test accounts created after the ban came into force was asked to verify their age, allowing them to remain active across major platforms. The findings suggest that the weakest point may not be age-verification technology itself, but the initial screening process, which relies heavily on self-declared ages and behavioural signals before triggering additional checks. While regulators argue that more sophisticated detection will improve over time, the results underscore how easily determined teenagers can circumvent age restrictions by entering false birth dates.

Meanwhile, a special panel convened by the European Commission has recommended restricting access to social media and other high-risk digital services for children under 13, arguing that platforms should prove they are safe before minors are allowed to use them. The panel’s recommendations leave room for member states to introduce additional precautionary measures for older adolescents if needed. Although the recommendations are not legally binding, they will inform future discussions on the Digital Services Act, the AI Act and wider EU child protection policies. President of the European Commission Ursula von der Leyen already confirmed that legislative proposals to strengthen children’s online protection will be presented after the summer.

At the same time, the EU’s approach has reignited a broader debate over where to draw the line between child protection and digital privacy. The European Parliament recently extended the bloc’s temporary ‘Chat Control 1.0’ framework until 2028, allowing technology companies to continue voluntarily scanning communications for known child sexual abuse material. Supporters argue that such measures provide authorities with an important tool to identify abuse and protect victims, while critics warn that automated scanning risks undermining privacy, secure communications, and encryption. Although the extension includes an amendment excluding end-to-end encrypted communications from scanning obligations, privacy advocates argue that questions around surveillance, false positives, and digital rights remain unresolved.

The glass slipper question. Cinderella had to leave the ball because the magic had a deadline. The digital world presents a different dilemma: the ball never ends. The digital world presents a different dilemma: the ball never ends. The examples above show that governments are likely to continue pursuing different paths—Britain with behavioural nudges, Australia with stronger restrictions, and the EU with greater platform responsibility—as they search for a solution that fits, or, to borrow from the fairy tale, the glass slipper.

IN OTHER NEWS LAST WEEK

Rope, Adult, Male, Man, Person, Face, Head

AI governance

Global governance. Twenty-nine countries signed an agreement in Shanghai on 16 July establishing the World Artificial Intelligence Cooperation Organization (WAICO), an independent intergovernmental international organisation that will be headquartered in Shanghai. At the conference, President Xi Jinping presented China as a partner for the Global South, calling for AI development through international cooperation rather than unilateral leadership. He announced 5,000 AI training and seminar opportunities for developing countries and pledged to deepen AI cooperation with regional organisations, including ASEAN, the League of Arab States, and the African Union.

Xi also called for AI systems to remain under human control and cautioned against using national security as a pretext to restrict AI development, in remarks widely seen as a response to US technology export controls. UN Secretary-General António Guterres welcomed WAICO as a natural evolution of China’s 2023 Global AI Governance Initiative, arguing that AI should be governed by all countries rather than a small group of states or companies. The organisation is intended to strengthen international cooperation on AI governance while addressing the technology’s economic, security, and ethical challenges.

Nobel laureates, scientists, religious leaders and former heads of state and government have signed the Rome Declaration for an Unarmed and Disarming Peace in the Age of Artificial Intelligence, Nuclear and Autonomous Weapons, New Digital Protocols, and Emerging Models of Digital Development. The declaration was adopted on 16 July on Rome’s Capitoline Hill following the Global Nobel Laureates Assembly on Artificial Intelligence and Nuclear War, hosted by the Vatican. The declaration calls for renewed international cooperation to address the challenges posed by AI, nuclear weapons and other emerging technologies. It stresses that decisions concerning life and death, peace and war, and the future of humanity should remain under meaningful human control, while highlighting the importance of ensuring that technological progress is guided by ethics, responsibility and respect for human dignity.

The USA. The White House has announced the launch of GOLD EAGLE, a cybersecurity vulnerability coordination initiative established under President Donald Trump’s Executive Order 14410, Promoting Advanced Artificial Intelligence Innovation and Security. According to the administration, the initiative brings together federal agencies, open-source software partners and operators of critical infrastructure to accelerate the identification and remediation of cybersecurity vulnerabilities using AI. GOLD EAGLE is intended to reduce duplicative vulnerability scanning, improve exploit detection and provide prioritised threat and remediation information to government and private-sector defenders.

Australia. The Australian government has announced plans for a national AI framework centred on new Australian AI Standards, enforceable rules for large AI data centres and stronger protections for creative works. Under the proposed standards, large AI data centres would be legally required to underwrite their additional power supply and pay the full cost of connecting to the electricity network. The government said this would prevent AI infrastructure expansion from increasing household energy bills. An Office of AI has been established within the Department of the Prime Minister and Cabinet to oversee the implementation of the Australian AI Standards. The framework will be considered by the National Cabinet in August, with legislation planned for early 2027.

South Korea. South Korea’s Ministry of Science and ICT has announced plans to launch the ‘AI for Everyone’ project this year, providing a homegrown AI service that anyone in the country can use free of charge without usage limits. The ministry will select participating companies through an open call for proposals. A beta version is scheduled for late September, followed by the launch of a general-purpose AI chatbot and an AI agent to help users search for and apply for public services.

Canada-South Korea. Canada and the Republic of Korea have signed a memorandum of understanding (MoU) between their respective Artificial Intelligence Safety Institutes (AISIs) to strengthen cooperation on AI safety and the governance of frontier AI systems. The partnership establishes a framework for exchanging information on AI technologies, emerging risks, testing methodologies, evaluation tools and governance approaches. The two institutes will also work together to advance internationally recognised methods for evaluating frontier AI models while identifying new areas for cooperation. A key element of the agreement focuses on risks associated with synthetic and AI-generated content.

Malaysia. Malaysia’s Ministry of Digital has launched public engagement sessions on its proposed AI Governance Bill with the aim to explain the proposed framework, encourage dialogue and gather feedback from government agencies, industry, businesses, academia and the public. It is intended to establish common principles that complement existing legislation and create a more coherent governance framework across sectors. The proposed legislation adopts a risk-based approach, recognising that AI-related risks can arise throughout a system’s lifecycle. Governance responsibilities would be allocated according to the roles and level of control exercised by different parties. The Ministry also highlighted incident reporting, appropriate safeguards and regulatory sandboxes as key mechanisms for supporting responsible innovation and strengthening public trust.

ITU. The International Telecommunication Union (ITU) has launched a new global initiative to develop international standards for trusted digital identity and agentic AI. Announced at the AI for Good Global Summit, the new Focus Group on Trust and Identity for Humans and Agentic AI will develop frameworks to strengthen accountability while ensuring meaningful human oversight of autonomous AI systems. The initiative will bring together experts from technology, policymaking, law and regulation to develop common terminology, reference architectures, trust frameworks, interoperability mechanisms and security benchmarks for AI agents. The group will also produce a roadmap to support future international standardisation of AI systems.

Digital self-reliance keeps momentum

Across different political systems and regions, governments are increasingly examining where their critical digital dependencies lie to ensure greater resilience in a world full of geopolitical tensions.

The Netherlands provides a clear example. Several national regulators, including the Authority for Consumers & Markets, the data protection authority, the Dutch central bank, and others, have warned that dependence on mainly US-based technology providers could have major consequences for businesses, governments, and society. They argue that digital autonomy does not mean complete independence, but rather ensuring freedom of choice through open standards, interoperability, data portability, and the ability to switch providers.

Switzerland is pursuing a similar discussion, with its armed forces reportedly reconsidering their reliance on Microsoft products for cyber units and exploring alternatives such as OpenStack. Italy, meanwhile, is focusing on AI capabilities, with its postal service announcing plans related to developing a national AI model tailored to local needs and contexts.

China’s approach, meanwhile, reflects a broader industrial strategy. President Xi Jinping has called for faster progress towards scientific and technological self-reliance, identifying AI, quantum technologies, advanced manufacturing, robotics, and semiconductors as strategic priorities. While emphasising innovation and competitiveness, China has also highlighted the need for stronger governance, ethical standards, and security measures for emerging technologies.

The EU has similarly been pursuing technological self-reliance through industrial policy and investment. The second phase of the European Tech Champions Initiative (ETCI), known as ETCI 2.0, has been launched to mobilise up to €80 billion for European technology scale-ups. The initiative seeks to strengthen Europe’s late-stage investment ecosystem by creating a pan-European platform to help highly innovative companies scale into global technology leaders—reflecting a broader effort to build domestic technological capacity and competitiveness.

Dutch government unveils plan to strengthen future workforce

The Dutch government has launched a national Talent Strategy to strengthen the country’s workforce and support future prosperity. The strategy aims to attract, educate, and retain talent in areas considered crucial to the Netherlands’ economic growth, public services, and response to social challenges.

The government identifies four priority domains for investment: digitalisation and AI; security and resilience; energy and climate technology; and life sciences and biotechnology. It says these are areas where the Netherlands has strengths in innovation and research, and where it wants to strengthen strategic autonomy.

Why does it matter? The approach also reflects a wider European challenge: countries need enough specialised workers for AI and emerging technologies, while also expanding lifelong learning so existing workers are not left behind.

UK establishes ministerial group to coordinate digital inclusion

The UK government has published the terms of reference for a new ministerial group created to coordinate digital inclusion policy across departments, aiming to improve access to digital services, technologies and skills.

The Ministerial Group for Digital Inclusion will set the government’s strategic direction, agree a shared vision and identify priorities for improving digital inclusion across the UK.

Its work will focus on embedding digital inclusion into policy design, public service delivery and existing government governance structures. The group will also monitor progress across departments with support from official-level bodies.

Why does it matter? Digital exclusion affects access to education, employment, healthcare and public services, making it a cross-government policy challenge rather than the responsibility of a single department. The new ministerial group is intended to improve coordination and ensure digital inclusion is considered more consistently across government.

View all updates

LOOKING AHEAD

Person, Face, Head, Accessories, Goggles, Binoculars

First substantive session of the UN Global Mechanism on cybersecurity

The first substantive session of the Global Mechanism on developments in the field of ICTs in the context of international security and advancing responsible State behaviour in the use of ICTs will be held on 20-24 July. This session marks the start of the substantive work of the Global Mechanism, a new single-track, permanent forum on ICT security under UN auspices.

LAST WEEK IN GENEVA

machine, Wheel, Spoke, City, Art, Bulldozer, Fun, Drawing

CERN scientists see AI shaping next-generation collider research

Meyrin, the western suburbs of Geneva, hosts the European Organization for Nuclear Research (CERN), the world-renowned particle physics research organisation behind the Large Hadron Collider, where researchers are expanding the use of AI across particle physics, deploying AI both during experiments and in the analysis of collision data. According to CERN physicists, AI systems are already being used to help identify potentially significant particle collisions in real time, enabling researchers to process the enormous volumes of data generated by the Large Hadron Collider. Scientists say AI’s role is expected to grow as CERN upgrades the LHC to the High-Luminosity LHC and develops plans for the Future Circular Collider (FCC), a proposed successor that is expected to begin operation in the 2040s. Beyond analysing experimental data, researchers are exploring how AI could support the design of detectors, optimise collider components and improve operational efficiency.

READING CORNER

The AI scramble for institutional knowledge – Diplo

As AI models harvest global data, a new geopolitical scramble is underway. Explore how nations can protect and negotiate sovereignty over their knowledge.

Geneva AI Week 2026: UN AI Governance Recap | Diplo

Diplo’s Executive Director Jovan Kurbalija recaps Geneva AI Week 2026, covering the Global AI Dialogue, AI for Good Summit, and WSIS Forum, plus the UN’s expanding role in AI governance.

DON’T MISS

Light, Outdoors, Nature, Sky, Scenery, Art, Night

Follow the road to the 2027 Geneva AI Summit, a global journey shaping AI’s future. Bookmark the page and register for updates to stay informed and involved every step of the way to Geneva 2027.

Weekly #270 Geneva takes centre stage in global AI governance

3 July – 10 July 2026

HIGHLIGHT OF THE WEEK

Geneva takes centre stage in global AI governance

Geneva has long been one of the places where global digital governance is translated into dialogue, institutions and practical cooperation. This week, that role became especially visible as WSIS Forum 2026, the inaugural Global Dialogue on AI Governance and AI for Good Global Summit 2026 brought governments, UN agencies, technical experts, companies, civil society and the research community into the same diplomatic ecosystem.

The concentration of events mattered because AI governance is reaching a point where principles are no longer enough. Countries are still facing very different levels of access to data, infrastructure, skills, compute, standards and regulatory capacity, while the technology is moving faster than many institutions can respond. Therefore, the matter is no longer whether AI should be safe, inclusive and human-centred, but how those commitments can be turned into working governance.

The timing matters. WSIS remains one of the longest-running multistakeholder processes in digital governance, rooted in the idea that digital technologies should support development, inclusion and public interest goals. Two decades after the original WSIS process began, the agenda is no longer only about connectivity or access to ICTs. It now includes AI, data governance, digital public infrastructure, online safety, labour markets, human rights, cybersecurity, digital divides and the future of the Internet Governance Forum.

At WSIS Forum 2026, that broader picture was visible across sessions on the IGF’s role in the WSIS era, youth mental health online, and AI’s impact on work and digitalisation. The discussion on the IGF focused on how a permanent mandate should translate into stronger government engagement, year-round collaboration and better links between global discussions and national and regional initiatives. Other WSIS sessions showed why that matters, linking digital governance to young people’s online lives, education, mental health, labour markets, skills and institutional adaptation.

The Global Dialogue on AI Governance added a different layer. Unlike WSIS, which has a long history of development and digital cooperation, the Global Dialogue is a newer mechanism created through the Global Digital Compact to support more inclusive international cooperation on AI. Its inaugural meeting in Geneva placed governments and stakeholders in direct conversation over AI’s opportunities, risks, divides, safety, trustworthiness and human rights implications.

The presentation of the first assessment by the Independent International Scientific Panel on AI was especially important because it brought an evidence-based layer into the political discussion. The panel warned that AI capabilities are advancing rapidly, while scientific understanding, regulatory systems and institutional capacity are struggling to keep pace. That gap affects who can evaluate AI systems, who can regulate them, who can benefit from them, and who may be exposed to risks without having enough voice in shaping the rules.

Why is this happening? AI governance is becoming too broad for any single forum, ministry, regulator or technical community to manage alone. It touches education, labour, health, defence, competition, human rights, energy, infrastructure, media, development and public administration. At the same time, countries do not enter the discussion from equal positions. Some are building frontier models and data centres, while others are still working to close connectivity gaps, develop local skills, secure public-sector capacity and participate meaningfully in global standard-setting.

That is why the AI divide is becoming one of the central questions in global governance. Access to AI is not only about whether people can use a chatbot. It is also about whether countries have the data, languages, compute, cloud access, regulatory expertise and institutional confidence needed to shape AI systems around their own priorities. If the global AI conversation focuses only on safety risks in advanced economies, it may miss the development, capacity and inclusion questions that determine whether AI reinforces old inequalities or helps reduce them.

AI for Good Global Summit 2026 provided the third part of Geneva’s AI week. While WSIS carried the development and multistakeholder legacy, and the Global Dialogue focused on AI governance at the UN level, AI for Good was built around practical applications, partnerships and the sustainable development goals. Its relevance lies in connecting AI governance to concrete use cases, standards, technical communities and implementation challenges.

What does this mean? The centre of gravity in AI governance is shifting from principles to institutional design. But, how to build mechanisms that make those principles operational: scientific assessments, capacity-building, standards, regulatory cooperation, accountability tools, rights protections, public-sector readiness and channels for countries with fewer resources to influence decisions.

Geneva’s strength has always been its diverse ecosystem: UN agencies, technical bodies, humanitarian organisations, diplomats, civil society, companies, standards experts and researchers. AI governance needs precisely that kind of ecosystem, because the technology does not fit neatly into one policy box. Yet bringing everyone into the room is only the first step. The real test is whether those conversations produce continuity, coordination and practical support after the summits end.

Wrapping up, the week showed why Geneva matters in the next phase of AI governance. WSIS connected AI to the long arc of digital development. The Global Dialogue placed AI risks, benefits and divides in a new UN process. AI for Good linked the debate to practical applications and the SDGs. Together, they showed that AI governance is no longer a side discussion within digital policy. It is becoming one of the main tests of whether global cooperation can respond to technological change without leaving most countries behind.

The question is whether Geneva can help turn a crowded AI governance conversation into a more coherent system of cooperation, accountability and practical support. If AI is becoming a general-purpose technology for economies, public services and societies, then global governance will be judged not by how many principles it produces, but by whether countries and communities have the tools, capacity and voice to shape how those principles are applied.

Diplo at Global Dialogue on AI governance, WSIS and AI for Good

Diplo maintained an active presence throughout the week at the Global Dialogue on AI Governance, the WSIS Forum, and the AI for Good Global Summit through high-level discussions, thematic sessions, side events, and exhibition booths that showcased its work on AI governance, digital diplomacy, and capacity development.

Diplo and GIP at the Global Dialogue on AI Governance

At the inaugural Global Dialogue on AI Governance, Dr Jovan Kurbalija, Executive Director of DiploFoundation and Head of the Geneva Internet Platform, co-chaired the thematic discussion on bridging AI divides, along with H.E. Samba Diouf, Minister of Communications, Telecommunications and Digital of Senegal. The session served as a structured multistakeholder exchange to share perspectives, experiences, and practical insights on bridging AI divides.

Diplo also co-organised a session with Coursera on Innovative approaches to future workforce development, where Diplo showcased its AI apprenticeship methodology, a hands-on approach that enables professionals in digital governance and diplomacy to build their own AI agents while exploring institutional questions such as knowledge ownership.

Diplo and GIP at WSIS Forum 2026

At WSIS Forum 2026, Diplo and the GIP co-organised a session with the Beijing Institute of Technology on Thursday, 9 July, exploring practical approaches for learning in the AI era. Drawing on both Chinese and European educational traditions, the discussion examined how AI is reshaping knowledge production, teaching, and learning, prompting a shift from education centred on knowledge transmission to mentoring, facilitation, and new curricula and assessment approaches. Participants also considered how AI apprenticeships and other learning-by-doing models can strengthen AI literacy by enabling learners to build and critically evaluate AI systems. The role of narrative, dialogue, and AI-supported personalised learning in fostering reflection, inquiry, and deeper understanding was also considered.

People, Person, Adult, Female, Woman, Crowd, Indoors, Male, Man, Accessories, Glasses, Audience, Bag, Handbag, Computer Hardware, Electronics, Hardware, Monitor, Screen, Elizabeth Quat, Mitch Dorge

Diplo and GIP also had a booth at the WSIS exhibition space, where we showcased publications and initiatives, including EspriTech de Genève and just-in-time reporting from the WSIS 2026 Forum through the Digital Watch observatory.

Desk, Furniture, Table, Adult, Female, Person, Woman, Male, Man, Accessories, Glasses, Standing Desk

Diplo and GIP at AI for Good Summit 2026

At AI for Good Summit 2026, Diplo and GIP participated in various sessions. Diplo’s Director of Partnerships, Tereza Horejsova, moderated one roundtable during the session, AI in education, research and skills development on Tuesday, 7 July 2025. The roundtable brought together high-level representatives of leading universities, industry actors, and international stakeholders to explore how AI is transforming higher education.

Diplo participated in the annual meeting of AI Skills Coalition partners, where the AI Skills Coalition community – representatives of international organisations, industry, academia, and the public sector – connected, shared priorities, and explored opportunities to scale impact through coordinated action.

Diplo and the GIP were also present at the Canton of Geneva space, showcasing our work on AI technology, policy, governance, and capacity development.

Advertisement, Poster, Accessories, Glasses, Adult, Female, Person, Woman, Male, Man, Wristwatch, Clothing, Formal Wear, Suit, Jewelry, Ring, Face, Head, Miha Mazzini, Dominique Cabrera, Peter Bennett-Jones, Amr Hamzawy

IN OTHER NEWS LAST WEEK

AI governance

The UN. UN Secretary-General António Guterres used the inaugural Global Dialogue on AI Governance in Geneva to call for global AI rules that are inclusive, safe and worthy of public trust. The discussions placed AI’s opportunities, risks, divides and human rights implications at the centre of the UN agenda, while the Independent International Scientific Panel on AI warned that scientific understanding and governance capacity are struggling to keep pace with rapid advances in AI capabilities.

Spain. Spain called for a UN-backed coalition to protect children from AI-related risks, warning that minors face growing exposure to synthetic content, manipulation, exploitation and harmful design features. The proposal, supported by France, the EU and Kenya, aims to bring governments, companies, civil society and international organisations together around stronger safeguards for children in the AI era.

Japan. Japan announced plans to establish a new AI council to drive national AI adoption and coordinate policy across government, industry and society. The initiative is expected to support AI use in public administration, healthcare, education and other sectors, while also addressing demographic pressures, workforce shortages and the need to review legal frameworks as AI becomes more widely deployed.

Portugal. Portugal launched AMALIA, its first open AI language model for European Portuguese, as part of a wider effort to strengthen linguistic and digital sovereignty. The model is designed to support public services, education, research and business use cases, while helping ensure that Portuguese-language users are not dependent only on systems trained primarily on larger global languages.

Germany. Germany launched an Agentic AI Hub for public administration, aiming to explore how autonomous and semi-autonomous AI systems could support government services. The initiative reflects growing interest in using AI agents to improve administrative efficiency, but also raises questions about oversight, accountability, procurement and the limits of automation in public decision-making.

European Commission launches AI cyber defence strategy

The European Commission presented an Action Plan on Cybersecurity and Artificial Intelligence to strengthen Europe’s response to AI-related cyber risks. The plan aims to help member states, businesses and public authorities use AI safely while addressing the risks created by advanced AI models.

The Commission said AI can help detect vulnerabilities, prevent cyberattacks and protect critical infrastructure. It also warned that malicious actors can use AI to automate attacks, identify weaknesses and carry out cyber operations at greater speed and scale.

The strategy focuses on the safe use of advanced AI, stronger EU cyber resilience and Europe’s AI capabilities for cybersecurity. The Commission will work with ENISA on a European Blueprint for secure access to advanced AI systems for cybersecurity purposes. A secure testing platform will support critical sectors, including energy, transport, health, finance, and public administration, while an EU Grand Challenge on AI for cybersecurity will support new AI-powered cybersecurity solutions.

Chinese AI models narrow the gap with US systems

Chinese AI models are moving closer to leading US systems in coding, reasoning and agent-based tasks, according to an analysis by the Center for Strategic and International Studies. CSIS said recent releases from Z.ai, Moonshot, DeepSeek and Alibaba-backed Qwen show that China’s rapid AI progress was not limited to DeepSeek-R1, but reflects a broader pattern of technical catch-up.

The analysis argues that Chinese models are now only months, rather than years, behind US frontier systems in several practical areas. It identifies knowledge distillation, open-weight research communities and efficiency-driven engineering as key factors behind this progress, while US chip export controls have pushed Chinese labs towards more efficient training and inference strategies.

Why does it matter? AI competition is no longer only about which country has the strongest frontier model. Chinese open-weight models are becoming capable, affordable and easier to deploy through third-party hosts or local infrastructure, which could shape global adoption among governments, startups and developers that cannot afford or do not want to depend entirely on US closed-model providers.

Europol denies bypassing EU data protection rules

Europol rejected allegations that it operated a ‘secret’ or ‘shadow’ database outside the EU data protection rules, saying recent reports misrepresented two long-established operational environments used to support digital investigations and online information analysis.

The agency said its Computer Forensic Network is used to analyse complex digital evidence securely. At the same time, its Internet-Facing Operational Environment collects and triages publicly available online information before relevant material is moved into Europol’s operational systems.

The dispute highlights the tension between law enforcement’s need to process large volumes of digital evidence and the privacy safeguards required under EU law. It also shows why transparency about investigative infrastructure matters for public trust, especially as law enforcement agencies modernise their data-processing capabilities for cybercrime, terrorism and serious organised crime investigations.

Illinois adopts AI safety law

Illinois Governor JB Pritzker signed the Artificial Intelligence Safety Measures Act, introducing new transparency and accountability requirements for large AI developers. The law requires qualifying developers to create risk mitigation frameworks, report potential large-scale harms and undergo annual independent third-party audits.

Why does it matter? The law shows how US states are moving ahead on AI safety in the absence of a comprehensive federal framework. By requiring independent audits, Illinois is pushing AI governance beyond company self-reporting and towards external verification. The approach could add pressure for national standards, but it may also deepen regulatory fragmentation if states continue to develop different AI safety rules.

Australia’s child social media ban faces enforcement delay

Australian Prime Minister Anthony Albanese criticised senators for delaying amendments to the country’s under-16 social media ban, after the proposal was referred to an eight-week Senate inquiry. The amendments would expand the powers of the eSafety Commissioner by allowing her to request documents, not only information, from platforms and third-party age-verification providers.

Why does it matter? The delay shows that child online safety rules increasingly depend on enforcement design, not only headline bans. Australia has moved further than many countries in restricting children’s access to major platforms. However, regulators still need evidence, platform cooperation and workable age-assurance mechanisms to turn legal obligations into practical protection.

View all updates

LOOKING AHEAD

First substantive session of the UN Global Mechanism on cybersecurity

The first substantive session of the Global Mechanism on developments in the field of ICTs in the context of international security and advancing responsible State behaviour in the use of…

High-Level Political Forum 2026

International Advisory Body for Submarine Cable Resilience

The International Advisory Body for Submarine Cable Resilience will hold its final meeting on 10 July (Friday). During the meeting, IAB Members will review and formally approve the final publication incorporating the reports and recommendations of the three Working Groups.

READING CORNER

What Geneva’s history can teach us about governing AI

Geneva’s centuries-old traditions of dialogue, inclusion, and compromise continue to offer valuable guidance for governing AI, as explored in a Diplo interview with Executive Director Jovan Kurbalija.

Claude Fable 5, frontier AI models and the future of cybersecurity

Digital resilience becomes increasingly important following Claude Fable 5 and its security implications.

DON’T MISS

Follow the road to the 2027 Geneva AI Summit, a global journey shaping AI’s future. Bookmark the page and register for updates to stay informed and involved every step of the way to Geneva 2027.

Weekly #269 From technology to the infrastructure stack

26 June – 3 July 2026

HIGHLIGHT OF THE WEEK

From technology to the infrastructure stack

For years, digital infrastructure was largely equated with connectivity: more fibre, faster broadband, and wider mobile coverage. But recent developments, including the ones from last week, show a broader shift in how governments think about infrastructure. It is no longer just about the networks that move data, but about the entire technological stack that enables the digital economy and society, from semiconductor supply chains and cloud platforms to satellite systems and quantum technologies.

Chips, cloud, connectivity, quantum and satellites have graduated out of the category of ‘technology’ and into the category of ‘infrastructure,’ which is a much older and more heavily governed category. It’s the same institutional maturing that electricity went through a century ago and highways went through half a century ago: useful and largely unregulated at first, then gradually recognised as too foundational to the rest of the economy to leave ungoverned.

Person, Adult, Male, Man, Book, Comics, Publication, Outdoors, Head

Canada and Germany’s new partnership on semiconductor supply chains reflects this evolution. Rather than focusing narrowly on access to chips, the agreement spans research, industrial development, investment, and innovation, signalling that semiconductor ecosystems are now seen as strategic infrastructure in their own right.

Singapore is taking a similar view of cloud infrastructure. Its proposed Digital Infrastructure Bill would bring cloud providers and data centres under a dedicated regulatory framework, recognising that the resilience of cloud services has become a matter of national interest rather than a purely commercial concern.

Elsewhere, governments continue to invest in the more familiar layers of the stack. Saudi Arabia’s rise to the top of global digital connectivity rankings underscores the continued importance of high-quality networks as the foundation of digital transformation. In South Africa, the launch of a state-backed satellite communications company highlights how governments are diversifying connectivity itself, adding new layers rather than relying solely on terrestrial infrastructure.

Some are already looking even further ahead. Türkiye’s first national quantum roadmap marks an early investment in technologies that may underpin future computing and communications, demonstrating that infrastructure planning is beginning to extend well beyond today’s commercially mature technologies.

At the same time, several governments are now explicitly building compute infrastructure as a national asset class. South Korea has announced a comprehensive AI infrastructure strategy centred on expanding high-performance computing capacity, including the procurement of 18,000 GPUs through a public-private national AI computing centre and a new national supercomputer. The strategy also couples infrastructure with capability-building: flagship LLM development, research access to compute and datasets, and talent pipelines through international AI labs and industry-university graduate programmes. Crucially, it extends infrastructure logic into deployment, mandating the rollout of domestically developed AI systems across core public services, from healthcare and education to disaster management and public administration.

Why does it matter? Digital infrastructure has always been about more than physical assets. Today’s policy agenda just makes that increasingly explicit. As governments rethink the foundations of their digital economies and societies, infrastructure is evolving from a collection of individual projects into a strategic architecture.

IN OTHER NEWS LAST WEEK

AI governance

The EU. The Council of the European Union has given its final approval to a regulation that simplifies parts of the EU’s AI framework as part of the broader ‘Omnibus VII’ package to reduce regulatory complexity. The updated rules revise the implementation timeline for high-risk AI systems, postponing full application until December 2027 for standalone systems and August 2028 for AI systems integrated into regulated products. The regulation also strengthens safeguards by explicitly prohibiting AI-generated non-consensual sexual content, including manipulated intimate imagery and AI-generated child sexual abuse material. Additional changes aim to reduce administrative burdens and improve legal clarity. Deadlines for establishing AI regulatory sandboxes have been extended to August 2027, transparency obligations for AI-generated content have been streamlined, and the regulation clarifies the division of responsibilities between EU and national authorities while reducing overlap with sector-specific legislation. The framework also clarifies the division of responsibilities between EU and national authorities and introduces mechanisms to avoid overlap with sector-specific legislation.

Malaysia. Malaysia has launched the Malaysia Digital Action Plan 2030 (MD2030), a national roadmap that places the Ministry of Digital at the centre of efforts to achieve the country’s ambition of becoming an AI-driven nation by 2030. The strategy aims to transform Malaysia from a consumer of technology into a producer of homegrown digital innovation through a coordinated, whole-of-government approach. The five-year plan sets national priorities across economic growth, digital public services, infrastructure, talent development, cybersecurity and AI innovation.

Japan. Japan’s industry ministry has approved ¥387.3 billion in funding for a domestic AI project to develop a multimodal foundation model for physical AI systems that control robots. The project aims to develop a foundation model that can be widely adopted by Japanese companies to support industrial automation and robotics. Officials see the initiative as a strategic effort to narrow the technology gap with the USA and China in next-generation AI.

The UK-Germany. The UK and Germany have agreed to strengthen cooperation on AI safety and security. The partners will deepen institutional cooperation by sharing best practices in AI evaluation, aligning research priorities and exchanging expertise. The collaboration will also examine the cybersecurity implications of advanced AI systems and contribute to the international evidence base on AI safety.

South Korea-Japan. South Korea and Japan have agreed to expand defence cooperation, following talks between South Korean Defence Minister Ahn Gyu-Back and Japanese Defence Minister Shinjiro Koizumi in Seoul. The agreement included a commitment to strengthen ties in state-of-the-art science and technology, including AI.

UNDP. UN Development Programme (UNDP) and the UN Department of Political and Peacebuilding Affairs (DPPA) have published a technical guide on AI in electoral administration. The publication, From Promise to Practice: AI in Electoral Administration, doesn’t endorse the use of AI for electoral processes; It provides a framework to help electoral authorities assess whether, where and how AI can be adopted responsibly.

Pax Silica members meet for second annual summit

Last Friday (26 June), the Second Pax Silica Summit, hosted by the USA, concluded. The initiative brings together a growing coalition of 24 countries and economies committed to building secure, resilient, and innovation-friendly AI ecosystems.

At the centre of the summit was a Joint Statement on AI Opportunity, signed by nearly three dozen partners including the USA, the EU, India, Japan, the UK and a wide range of economies across Latin America, the Middle East, and Asia. The statement signals a shared preference for a pro-growth and innovation-driven approach to AI governance, emphasising support for startups, developers, and private-sector dynamism, while also strengthening trust and security across global AI supply chains.

The summit also moved beyond principles into implementation. A pilot project in Panama will test an AI-enabled logistics and customs platform designed to track and verify sensitive goods such as semiconductors, critical minerals, and AI infrastructure components. The aim is to speed up trade for trusted shipments while improving oversight of strategic technologies. If successful, the system could be expanded across other Pax Silica members.

In parallel, the USA and Stanford University launched the “Foundry School” initiative, an education and workforce programme focused on advanced manufacturing. It combines executive-level seminars with a transferable curriculum designed to help partner countries build domestic expertise in industrial systems that underpin technological competitiveness.

Why does it matter? Having expanded by 10 members at its second summit, Pax Silica points to the gradual formation of a technology-aligned geopolitical bloc, where influence is exercised less through traditional diplomacy and more through control over industrial ecosystems, trade infrastructure, and standards-setting coalitions.

Singapore updates its cybersecurity playbook

Singapore has unveiled two initiatives that strengthen its approach to governing digital risks across multiple fronts.

Singapore’s Online Safety Commission has begun operations, giving victims of online harms a dedicated channel to seek faster support and redress. In its first phase, the commission will support victims affected by five categories of online harm: online harassment, including online sexual harassment, doxxing, online stalking, intimate image abuse and image-based child abuse.

Victims of online harassment and online stalking are generally expected to report harmful content to the relevant platform first. If the platform fails to respond promptly or provides an inadequate response within 24 hours, the platform may be reported to the commission. More serious harms, including doxxing and image-based abuse, can be reported directly.

Where there is reason to suspect that online harm has occurred, the Commissioner may issue directions to the person who posted the content, the administrator of the online space or the platform hosting it. These directions may require access to harmful content to be disabled or an account to be restricted. Non-compliance is a criminal offence.

Singapore is also introducing court-based remedies through statutory torts. Victims may bring civil claims against communicators, administrators, or platforms that fail to meet the duties set out in the law. For intimate image abuse and image-based child abuse, courts must award at least $5,000 for each image or recording if the claim succeeds.

The commission will also work with community partners that can provide counselling and practical support to victims and families.

Separately, Singapore’s Cyber Security Agency (CSA) has outlined a broad package of initiatives to strengthen cybersecurity as AI reshapes the threat landscape. According to the Singapore Cyber Landscape 2025/2026 report, AI is enabling faster and more sophisticated cyberattacks, with agentic AI capable of automating parts of the cyber kill chain and accelerating vulnerability research, exploit development, and attack preparation. At the same time, AI is improving cyber defence by enhancing threat detection, incident response, and vulnerability management, making the secure deployment of AI systems an increasing priority.

To support this, CSA has published guidelines for securing AI systems, developed guidance on agentic AI security, and is working with international partners on AI security standards. The agency also warned that AI is making phishing and scams more convincing through voice cloning, deepfakes, and personalised attacks, prompting initiatives such as a national simulated scam exercise to improve public awareness.

Beyond AI, the report highlights a sharp rise in infected infrastructure, continued ransomware threats—particularly for SMEs—and growing risks from insecure consumer IoT devices. Singapore is responding by tightening cybersecurity requirements for residential routers and critical infrastructure operators, supporting SMEs through advisory and recovery programmes, advancing its National Quantum-Safe initiative, and expanding cybersecurity capabilities through exercises, partnerships, and new certification requirements. Together, the measures reflect Singapore’s effort to prepare for a threat environment which is increasingly shaped by AI, critical infrastructure risks, and future quantum technologies.

Why does it matter? As AI accelerates both online harms and cyber threats, policymakers are complementing regulation with new institutions, operational capabilities, technical standards, and support mechanisms.

CJEU court upholds €4.7 billion fine against Google in Android case

The Court of Justice of the European Union (CJEU) has upheld a €4.1 billion ($4.67 billion) antitrust fine against Google, rejecting the company’s appeal in a long-running case over its Android mobile operating system.

The European Commission imposed the record fine in 2018 after finding that Google had abused its dominant position by requiring smartphone manufacturers to pre-install Google apps and imposing contractual restrictions that limited competition. The ECJ dismissed Google’s appeal, confirming the Commission’s findings.

A lower EU court reduced the penalty from €4.34 billion to €4.1 billion in 2022 while largely upholding the Commission’s decision. Google has argued that Android increases consumer choice and supports developers, and said it modified its contractual arrangements following the original ruling to comply with EU competition rules.

The judgement comes as the EU continues to intensify scrutiny of major technology companies through both competition law and the Digital Markets Act, alongside investigations into digital advertising and other platform practices.

Why does it matter? The ruling reinforces the EU’s long-standing approach to digital competition, confirming that dominant technology companies can face substantial penalties when their market position is used to restrict consumer choice or limit competition. It also strengthens the Commission’s record in pursuing landmark antitrust cases against major digital platforms.

EU introduces €3 duty on low-value e-commerce imports

The European Commission introduced a temporary €3 customs duty on low-value goods imported into the EU from outside the bloc, from Wednesday, 1 July 2026. The duty applies to a wide range of commonly purchased goods, including clothing, toys, and electronics, covering items worth up to €150.

The duty is charged per customs tariff classification rather than by quantity. For example, purchasing five T-shirts attracts a single €3 charge because they share the same tariff code, whereas buying three T-shirts and a watch incurs two €3 charges because they fall under different classifications. Sellers or importers will declare and pay the duty through the customs process.

Why does it matter? The measure is intended to create fairer competition for the EU businesses, improve consumer protection by strengthening oversight of imported goods, reduce customs fraud linked to undervaluation and false declarations, and address the environmental impact of growing volumes of low-value shipments. It is also widely seen as a measure for discouraging shopping via low-cost e-commerce platforms.

Mauritius unveils fintech strategy to boost digital finance growth

Mauritius has launched its National Fintech Strategy 2026–2030, a roadmap aimed at strengthening digital finance, innovation and financial inclusion.

The roadmap focuses on six areas: regulation and innovation, digital infrastructure and cybersecurity, skills development, market growth, international cooperation and consumer protection.

Implementation will run until 2030 and will be overseen through a dedicated governance framework. Planned targets include shorter licensing approval times, expanded digital onboarding, stronger digital infrastructure and training more than 5,000 people each year in specialised fintech skills.

The government says the strategy is intended to position Mauritius as Africa’s trusted fintech hub while supporting sustainable growth and the wider digital transformation of financial services.

Why does it matter? Mauritius’ strategy reflects a wider African policy trend: governments are trying to move fintech from fragmented innovation into structured national development plans.

View all updates

LAST WEEK IN GENEVA

The 62nd session of the HRC continues

The 62nd session of the UN Human Rights Council (HRC) is ongoing until 7 July. Last Friday, Peggy Hicks, Director, Thematic Engagement, Special Procedures and Right to Development Division of the Office of the United Nations High Commissioner for Human Rights, presented seven reports to the Human Rights Council. Two of them are related to digital technology.

The Report of the UN High Commissioner for Human Rights on Using digital tools for stakeholder engagement: aligning corporate practice with the Guiding Principles on Business and Human Rights (A/HRC/62/25) examines how companies are increasingly using digital tools to engage with workers, communities, and other affected stakeholders in the context of the UN Guiding Principles on Business and Human Rights. It finds that while these tools can improve reach and responsiveness, they can also deepen exclusion, weaken trust, and raise privacy concerns when poorly designed or used without safeguards. Importantly, digital engagement is not automatically meaningful—its value depends on whether it is appropriate, accessible, and rights-respecting for the people involved.

The report stresses that digital tools should complement, not replace, direct engagement with stakeholders such as workers, unions, and communities. It provides guidance for companies, technology providers, regulators, and civil society on assessing whether such tools are suitable, safe, and genuinely contribute to meaningful participation. Ultimately, it argues that digital engagement is only as rights-respecting as the design choices and governance frameworks behind it.

The Study on States’ obligations under international human rights law, relevant norms and commitments, and the human rights responsibilities of business enterprises in line with the Guiding Principles on Business and Human Rights, across the lifecycle of new and emerging digital technologies (A/HRC/62/33) frames human rights due diligence as a continuous process to identifying, preventing, mitigating, and remedying harms, triggered by actual violations as well as foreseeable risks. States are expected both to avoid rights violations in their own use of AI and to regulate private actors through robust legal frameworks, oversight, and, where necessary, restrictions on high-risk applications.

UN scientific panel publishes first global AI assessment

The UN Independent International Scientific Panel on Artificial Intelligence has published its first preliminary report, providing an evidence-based assessment of AI’s opportunities, risks, and societal impacts.

The report describes AI as a general-purpose technology capable of boosting productivity, innovation, and access to services. However, it stresses that these benefits are unevenly distributed, with significant gaps in access to computing power, talent, infrastructure, and data. Developing countries risk becoming consumers rather than producers of AI unless they invest in local ecosystems, digital skills, and research capacity, including multilingual systems and locally relevant datasets.

Alongside opportunities, the report identifies growing risks, including AI-enabled cyberattacks, fraud, disinformation, election interference, and market concentration. It also highlights the lack of technical capacity in many governments to independently assess frontier AI systems. Environmental impacts are also flagged, particularly rising energy, water, and hardware demands linked to large-scale AI deployment.

The report will serve as a key input to the Global Dialogue on AI Governance, which takes place on 6–7 July alongside the World Summit on the Information Society (WSIS) Forum and the AI for Good Global Summit in Geneva.

LOOKING AHEAD

Global Dialogue on AI Governance inaugural meeting

The Global Dialogue on AI Governance (AI Dialogue) will hold its inaugural meeting on 6-7 July 2026. The programme of the Global Dialogue will include high-level governmental plenary segments, multistakeholder exchanges, and the presentation of the report of the multidisciplinary Independent International Scientific Panel on Artificial Intelligence, thematic discussions across four clusters, and side events.

WSIS Forum 2026

The 2026 edition of WSIS Forum will convene representatives from governments, international organisations, civil society, the private sector, academia, and technical communities to exchange knowledge, highlight innovative solutions, and engage in multistakeholder dialogue to advance the implementation of WSIS Action Lines and contribute to broader global development goals.

AI for Good Global Summit 2026

The AI for Good Global Summit 2026 starts on Tuesday, 7 July, in Geneva. The summit will focus on exploring and advancing the practical application of AI to address global challenges and support the SDGs. The programme includes keynote speeches, panels, exhibitions, workshops, and related initiatives to identify scalable AI applications that contribute to sustainable development and societal benefit. The event also offers opportunities for networking, knowledge exchange, and stakeholder engagement across the global AI ecosystem.

High-Level Political Forum 2026

The High-level Political Forum on Sustainable Development 2026 (HLPF 2026) will convene from 7 to 16 July. Discussions will place digital technologies, including AI, at the centre of SDG 9 and broader development strategies, highlighting their role in transforming industry, innovation systems, and global value chains. At the same time, they stress persistent digital divides, with uneven access to connectivity, computing capacity, and digital infrastructure—particularly in LDCs, LLDCs and SIDS

International Advisory Body for Submarine Cable Resilience

READING CORNER

Digital Watch newsletter – Issue 111 – June 2026

In our June 2026 issue, we examine Geneva’s role as a central hub for global AI governance, where diplomacy, technical standards, and human-centred approaches increasingly converge. We also look at how Anthropic’s brief suspension and reinstatement of Fable 5 reflects shifting US export controls and increasingly reactive AI regulation. We explore Africa’s efforts to strengthen its voice in global digital diplomacy amid coordination gaps, alongside the European Commission’s Technological Sovereignty Package aimed at boosting capabilities in AI, semiconductors, and cloud infrastructure. We also introduce a new structured snapshot of AI developments across policy, industry, and governance.

Women and AI: Reflecting bias or reinforcing inequality?

When AI learns from humanity, it inherits more than knowledge. It also absorbs the biases, assumptions, and inequalities embedded in society.

DON’T MISS

Follow the road to the 2027 Geneva AI Summit, a global journey shaping AI’s future. Bookmark the page and register for updates to stay informed and involved every step of the way to Geneva 2027.

Digital Watch newsletter – Issue 111 – June 2026

June 2026 in retrospect

Geneva accounts for an estimated half of global AI governance discussions, reflecting its role as a central node in the international policy landscape. We analyse how this institutional density is transforming the city into a space where technical standards, diplomacy, and human-centred governance intersect, and where the foundations of emerging global AI governance frameworks are increasingly negotiated.

Across the Atlantic, Anthropic briefly suspended and then reinstated access to Fable 5 within a span of 15 days, in line with evolving US export control requirements. We explore how such reactive regulatory moves in the AI space can shape the broader ecosystem, and what they reveal about the wider governance challenges posed by rapidly advancing AI systems.

Beyond this, AI developments continue to accelerate across policy, industry, and governance. In this edition, we are also experimenting with a new way of visualising AI-related developments, offering a more structured, accessible overview of key developments throughout the month.

As AI, cybersecurity, and data-driven technologies become embedded in global governance, Africa is seeking to strengthen its influence in shaping emerging digital rules, but fragmented regional coordination and uneven implementation continue to limit the coherence of its voice. We look at how bridging continental strategies with national action could determine whether Africa moves from participation to meaningful agenda-setting in global digital diplomacy.

The European Commission has presented its long-awaited European Technological Sovereignty Package to strengthen Europe’s capacity in semiconductors, AI, cloud infrastructure, and open-source technologies. We examine the package’s key building blocks and the broader geopolitical and industrial context shaping Europe’s bid for greater technological autonomy.

Our snapshot of broader headlines goes beyond AI, with cybersecurity, technological and infrastructure developments dominating the agenda.

AI governance in June

Geneva at the centre of AI governance: Where technology, diplomacy, and humanity converge

Around half of the world’s AI governance discussions take place in Geneva. While the figure is necessarily an estimate, it reflects the extraordinary concentration of international organisations, diplomatic missions, standards bodies, and policy processes located within the city’s international district.

Few places bring together so many institutions responsible for shaping how AI is developed, deployed, and governed across such a wide range of issues—from technical standards and telecommunications to trade, labour, human rights, humanitarian action, and international peace and security.

As AI rapidly transforms economies, societies, and geopolitics, Geneva has become much more than a venue for international meetings. It is emerging as the world’s principal hub for AI governance, where governments negotiate new norms, international organisations develop policy frameworks, technical experts establish standards, and diverse stakeholders work to translate broad principles into practical governance. Increasingly, the city’s influence lies not in any single institution, but in the interactions among them, creating an ecosystem where the technological, political, economic, and human dimensions of AI can be addressed together.

A dense ecosystem of global governance

Over decades, Geneva has become home to institutions that collectively span the full governance spectrum of digital technologies: trade rules, telecommunications infrastructure, technical standards, humanitarian coordination, labour rights, human rights protection, and international security. In the context of AI, this institutional density is becoming particularly consequential, as governance challenges extend simultaneously across economic, social, ethical, legal, and geopolitical domains.

At the centre of Geneva’s role in digital governance are organisations whose mandates predate the AI era but now directly shape its trajectory.

The World Trade Organization (WTO) influences the global framework governing trade, data flows, digital services, and the supply chains underpinning AI development and deployment. Its discussions shape the economic environment in which AI technologies circulate globally.

The International Telecommunication Union (ITU), as the UN specialised agency for digital technologies, provides one of the key global forums for technical cooperation on connectivity, digital infrastructure, and emerging technologies, including AI. Its role increasingly extends into bridging technical standards, policy dialogue, and development priorities.

Alongside these institutions, Geneva hosts key global standard-setting bodies such as the International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC). While often operating behind the scenes, these organisations define the interoperability, safety, and compatibility frameworks that allow digital systems—including AI systems—to function across borders and sectors. As AI becomes more embedded in critical infrastructure and public services, the significance of such standards increases accordingly.

This institutional landscape is complemented by Geneva’s diplomatic ecosystem, where nearly every UN member state maintains representation. The result is a continuous presence of diplomatic engagement across multiple policy domains, enabling AI governance discussions to take place not only in formal negotiations but also in sustained informal exchanges.

Between technology, diplomacy, and intellectual tradition

Geneva’s role in AI governance is not only institutional but also intellectual and historical. The city sits at the intersection of technological development and longstanding philosophical inquiry into the relationship between knowledge, power, and responsibility.

One of the most frequently cited cultural references in this context is Mary Shelley’s Frankenstein, written near Lake Geneva in 1816. Often regarded as an early reflection on scientific responsibility, the novel raises questions about creation without control, unintended consequences, and moral accountability—questions that resonate strongly in contemporary debates about increasingly autonomous AI systems.

Similarly, the literary work of Jorge Luis Borges, particularly his explorations of infinite information, classification, and human cognition, has become a useful conceptual reference point in an era defined by large-scale data processing, generative systems, and algorithmic production of knowledge.

Together, these intellectual traditions provide a broader lens through which AI governance debates in Geneva are often interpreted: not only as technical or regulatory questions, but as continuations of longstanding inquiries into the limits of knowledge, control, and human agency in the face of complex systems.

Art, Painting, Face, Head, Person, Photography, Portrait, Voltaire, Jean-Jacques Rousseau, Valentin Haüy

Geneva as the centre for AI diplomacy

Beyond its historical and institutional significance, Geneva has become an increasingly active venue for international discussions on AI governance. The city hosts a growing number of meetings, conferences, and policy dialogues dedicated to the governance of AI and other emerging technologies. Geneva also hosts a range of other initiatives focused on AI governance, including policy dialogues, expert consultations, and multistakeholder discussions addressing issues such as human rights, health, humanitarian action, sustainable development, trade, and technical standards.

Take, for instance, June of 2026. At the 114th International Labour Conference, digital labour governance was firmly placed on the international agenda. Governments, employers, and workers from the ILO’s 187 member states engaged in negotiations on platform work, including a draft Convention supplemented by a Recommendation on decent work in the platform economy. These discussions reflected growing recognition that AI and platform-based systems are reshaping labour markets, employment relationships, and social protection frameworks.

Alongside these negotiations, the ILO Director-General’s report, A moment of choice: Harnessing AI for decent work, framed AI as more than a productivity tool. It positioned AI as a structural governance challenge affecting jobs, skills, inequality, institutional adaptation, and social dialogue. The combined focus on platform work and AI illustrated how labour governance is increasingly inseparable from questions of digital transformation.

Shortly afterwards, attention shifted from labour to international peace and security. UNODA convened informal exchanges on AI in the military domain, responding to a UN General Assembly mandate to explore emerging normative frameworks, risks, and opportunities. The discussions brought together states, international organisations, civil society, academia, industry, and technical experts.

The same week, UNIDIR convened its Global Conference on AI, Security and Ethics 2026, bringing together diplomats, policymakers, researchers, and military experts to examine how AI is reshaping global security environments. Discussions spanned algorithmic bias, dual-use technologies, AI assurance, cyber defence applications, infrastructure dependencies, and public–private responsibility chains.

During the conference, UNIDIR also launched a Centre of Excellence on AI, Peace and Security. Designed as a permanent platform, the centre aims to address fragmentation in global AI governance by consolidating expertise, facilitating multistakeholder dialogue, and supporting the translation of high-level principles into operational governance approaches. Its model combines evidence generation, inclusive dialogue, and implementation-oriented pathways, positioning it as a coordination space rather than a competing initiative.

Book, Comics, Publication, Adult, Male, Man, Person, Face, Head

July 2026 further exemplifies Geneva’s role in AI governance. The inaugural Global Dialogue on AI Governance will convene governments and stakeholders on 6 and 7 July to advance international cooperation on AI. Its mandate is explicitly broad: to facilitate the exchange of best practices, support inclusive discussion, and ensure that AI contributes to the SDGs while helping to bridge digital divides within and between countries.

The Dialogue is structured around four thematic clusters: AI opportunities and implications across social, economic, ethical, cultural, linguistic, and technical dimensions; bridging AI divides through capacity-building and infrastructure development; ensuring safe, secure, and trustworthy AI through interoperability and compatibility of approaches; and strengthening respect for human rights through transparency, accountability, and human oversight.

Its multistakeholder design reflects Geneva’s broader governance philosophy. Participation includes UN member states, international organisations, UN agencies, and accredited representatives from civil society, academia, industry, and technical communities. Importantly, the dialogue is positioned not as a standalone negotiation but as part of a wider ecosystem of events taking place in the same week.

Running alongside it is the AI for Good Global Summit, organised by ITU in partnership with more than 50 UN agencies and co-convened with Switzerland. The summit focuses on the practical application of AI to global challenges, including sustainable development, capacity-building, standards development, and ethical considerations. It brings together governments, international organisations, industry, academia, and technical communities through a mix of panels, exhibitions, workshops, and demonstrations.

These events show that it is in Geneva that questions of standards, security, development, labour, ethics and human rights increasingly inform one another.

Beyond events: An emerging governance architecture

Looking further ahead, this trajectory is expected to culminate in the 2027 Geneva AI Summit. More than a standalone conference, the summit is intended to build on the growing ecosystem of institutions, dialogues and governance processes that are taking shape across Geneva and globally.

As governments, international organisations, and societies continue to grapple with the opportunities and risks associated with AI, Geneva’s role as a centre for digital governance is likely to become increasingly significant.

The city’s unique combination of technical expertise, standards-setting institutions, diplomatic networks, and human-centred governance traditions provides a platform for addressing complex questions that no single actor or sector can solve alone.

This text is an adaptation of the Diplo interview ’Is global AI governance shaped in Geneva?’ with Dr Jovan Kurbalija. A deeper dive into how Geneva’s intellectual and diplomatic heritage informs today’s challenges in AI governance can be found in Kurbalija’s interview, ‘Esprit de Genève in the AI era’, which examines the works of Mary Shelley, Jorge Luis Borges, Rousseau, Charles Bonnet, and Ferdinand de Saussure, linking literature, philosophy, linguistics, and early scientific thought to contemporary questions about AI.

Don’t miss!

Lighting, Light, Pattern, Purple, Art, Modern Art, Accessories, Ornament, Graphics

Follow the road to the 2027 Geneva AI Summit, a global journey shaping AI’s future. Bookmark the page and register for updates to stay informed and involved every step of the way to Geneva 2027.

The Fable 5 case and the problem of reactive AI governance

Anthropic’s release of Claude Fable 5 was meant to mark a milestone: the first publicly available model from its powerful Mythos family, launched across Claude, the API, AWS Bedrock, Google Vertex AI, and Microsoft Foundry. The company presented Fable 5 as offering the capabilities of Mythos 5 while incorporating safety measures suitable for public deployment.

The milestone proved short-lived. Just three days after the launch, on 12 June, the US government ordered Anthropic to suspend access to both Fable 5 and Mythos 5 for foreign nationals inside and outside the USA. Anthropic suspended access to both models for all users because it said it had no reliable way to verify nationality in real time. By the end of the day, both models had been taken offline globally. And just like that, a major product launch turned into a striking illustration of the growing influence of governments over frontier AI deployment.

A jailbreak vs normal behaviour. According to reports, the directive was triggered by a demonstration in which Fable 5 read a codebase and fixed vulnerabilities. US officials interpreted this as the model bypassing safety controls, a ‘jailbreak‘ that could let attackers use the same capabilities to compromise systems, steal data, or cause harm.

Anthropic, on the other hand, claims that the government misunderstood the demonstration. The vulnerabilities were already known and minor. Other models, including GPT-5.5, can find the same vulnerabilities without any safety bypass. This wasn’t a jailbreak, but normal model behaviour that was misinterpreted.

Regardless of who was right, this development reveals a structural gap: There is no established process for the evaluation of high-capability models prior to release. The government sees something alarming and acts. The company says, ‘Wait, you misunderstood ‘. But by then, the model is already shut down.

Reactive regulation. The Fable 5 case is a textbook example of reactive regulation: Triggered by an alarm, not by a pre-existing framework; No prior criteria for evaluating when Mythos-class models should be restricted; Punitive first response: shut it down before understanding the situation; A communication gap leads to distrust.

Why does this matter? Because reactive regulation is unpredictable. Companies can’t plan their releases. The public gets whiplash. Innovation suffers from uncertainty.

Person, Electronics, Hardware, Face, Head, Computer Hardware, Computer

The cost of governing by alarm. The current administration has been vocal about not wanting to drown AI innovation with regulation and has even blocked some state-level AI regulation attempts. But Fable 5 shows that when capability crosses into security concerns, the government doesn’t need legislation. It uses export controls.

Export controls are not regulation in the traditional sense. They operate under different legal authority, namely national security. The administration can still say it is not regulating AI while using security powers to control it.

Fable 5’s suspension was, ultimately, temporary. Antrhopic added a targeted safety classifier that blocks the specific cybersecurity jailbreak technique originally flagged by Amazon researchers. Instead of altering the core model, this filter stops the bypass mechanism while automatically redirecting flagged requests to Claude Opus 4.8 to safely handle complex coding tasks. To maintain security, the deployment enforces a mandatory 30-day data retention window for continuous monitoring and adjusts billing so users are not charged premium rates if their prompts are rerouted. The export controls were lifted on 30 June.

But while it may seem that the storm has blown over, there may have been a lasting impact on future policy moves: Export controls may very well become the default AI policy tool in the USA.

What does this mean for future model releases? Companies may hesitate to release powerful models publicly if they are at risk of sudden shutdown. Governments may issue directives more quickly and with less deliberation if the threshold for alarm is low. Public access to transformative tech becomes more uncertain. The tension between innovation and security intensifies as models get more capable.

The global dimension matters too. Other countries may follow the US export control approach. If the USA uses security directives to shut down models, other governments may do the same for their own reasons.

What does this mean for humanity? When governments shut down powerful models reactively, it affects researchers studying frontier AI, companies building on the models, and users who have only just gained access to some of the most capable AI systems available.

The tension is between AI as a public good, meaning open access, democratised knowledge, broad benefit, and AI as a security risk, meaning restricted, controlled, and kept from the public.

At its core, the question is whether humanity’s most powerful technologies should remain in public hands or be locked away when they provoke concern.

This text is an adaptation of Slobodan Kovrlija’s blog ‘Fable 5 and the gatekeeping of human reasoning.’

www.diplomacy.edu

Fable 5 and the gatekeeping of human reasoning – Diplo

Anthropic’s Claude Fable 5 went public, then vanished 72 hours later. Inside the sudden U.S. government shutdown and what it means for the future of AI power.

Africa’s digital diplomacy: Building a common voice for global digital governance

Africa’s role in global digital governance is expanding alongside AI, cybersecurity, and data-driven technologies. As these systems become embedded in development, trade, security, and public services, questions of representation, digital sovereignty, institutional capacity, and regional coordination are increasingly central. A key challenge is how African states can shape emerging global rules on AI governance, cybersecurity, telecommunications, and internet governance while ensuring domestic priorities are reflected.

Africa’s place in an evolving digital governance landscape. Digital technologies now cut across core areas of statecraft, including trade, education, healthcare, humanitarian response, and national security. As a result, digital governance has become a central pillar of contemporary diplomacy rather than a technical niche.

African countries are actively engaged in global digital policy processes through multilateral institutions and regional frameworks. However, concerns remain about the consistency and depth of representation, especially as negotiations on AI governance and cybersecurity intensify. Upcoming global processes on AI and internet governance are seen as key opportunities where coordinated African positions could increase influence.

Cyber diplomacy evolving into broader digital diplomacy. The distinctions between cyber diplomacy, digital diplomacy, and technology diplomacy are increasingly blurred. These fields are converging into a single diplomatic domain covering AI systems, data flows, digital infrastructure, and cyber stability.

Diplomacy in this context is defined less by terminology and more by the ability of states to influence global norms and technical standards. African participation in AI governance, cybersecurity, and internet governance is growing, but its effectiveness depends on coordination and the ability to translate engagement into negotiated outcomes.

Regional coordination remains uneven. Africa participates in global digital governance through AU-led frameworks and regional organisations such as ECOWAS, SADC, EAC, and COMESA. Despite this institutional architecture, translating continental strategies into national policy remains inconsistent.

Frameworks such as the AU digital transformation agenda and cybersecurity instruments like the Malabo Convention exist, but awareness and implementation vary widely. Many states develop independent regulatory approaches to data protection and cybersecurity, often without fully integrating continental standards.

Limited domestic financing for regional initiatives further constrains implementation, reinforcing reliance on external support and weakening policy ownership.

Strengthening coordination through shared positions. Efforts to improve African coordination in digital negotiations are gradually increasing. In telecommunications, structured mechanisms are being used to align national positions ahead of international standard-setting processes, particularly within the International Telecommunication Union.

These mechanisms aim to reduce fragmentation by encouraging shared priorities and common negotiating positions. Similar coordination is emerging in digital trade, where regional initiatives support electronic documentation, customs digitisation, and cross-border interoperability.

However, progress is uneven. Cybersecurity and cybercrime cooperation advance more slowly due to political sensitivity and the complexity of harmonising legal frameworks. Smaller regional groupings often move faster due to fewer actors and more focused agendas.

Persistent challenges in implementation. Implementation remains a structural weakness in African digital governance. While continental frameworks are increasingly well developed, national adoption is inconsistent.

Key drivers include capacity constraints, institutional fragmentation, and dependence on external funding. These factors limit ownership and slow the integration of continental priorities into domestic legal systems, resulting in uneven uptake across regions and policy areas.

Building a more coherent African digital agenda. There is growing recognition of the need for stronger continental coordination in digital governance. This includes aligning positions in global negotiations, improving coherence between regional and national policies, and strengthening institutional and technical capacity.

A more unified approach is seen as essential for ensuring that African priorities are consistently reflected in global debates on AI governance, cybersecurity norms, and digital economic regulation. This requires not only engagement in international forums but sustained investment in expertise and implementation capacity at home.

Looking ahead. Africa’s influence in global digital governance will depend on its ability to convert participation into coordinated positions and reduce gaps between continental ambition and national implementation. As global discussions on AI, cybersecurity, and digital cooperation intensify, internal coherence will become increasingly important.

Digital governance is now closely linked to broader development priorities, including infrastructure, education, and economic policy. Strengthening Africa’s role therefore requires integrated approaches that connect policy design, regional coordination, and implementation across governance levels.

Ultimately, Africa’s digital diplomacy will be defined not just by its presence in global negotiations, but by its capacity to act as a coherent and strategically aligned actor in an evolving digital order.

This text is informed by Diplo’s webinar ‘Cyber Diplomacy in Africa: Regional, National and Continental Initiatives.’ Watch the event recording below.

‘Tech Liberation Day’: EU launches tech sovereignty push

The European Commission has presented its long-awaited European Technological Sovereignty Package to strengthen Europe’s capacity in semiconductors, AI, cloud infrastructure, and open-source technologies. We examine what tech sovereignty is, what it is not, the building blocks of the package, and the context behind it.

What is technological sovereignty? A 2025 JRC report defines it as ‘Europe’s ability to develop, control and scale the critical technologies, infrastructure, services and data, including digital ecosystems, that underpin its economy, security and society, while derisking and diversifying supply chains and technological exposure to reduce strategic dependencies and resist foreign interference.’

What tech sovereignty isn’t. The Commission was explicit: Tech sovereignty is not isolation, protectionism, or tech decoupling. A technologically sovereign EU would remain integrated into the global economy but would better manage risks and opportunities stemming from global technological interdependence.

The building blocks. The package brings together two legislative proposals—the Chips Act 2.0 and the Cloud and AI Development Act (CADA)—alongside an Open Source Strategy and a Strategic roadmap for digitalisation and AI in energy.

Book, Comics, Publication, Adult, Male, Man, Person, City, Face, Head, Accessories, Bag, Handbag, Art, Road, Street, Urban

Chips, chips, chips. The first pillar of the package focuses on semiconductors. The original Chips Act, adopted in 2023, was Europe’s response to the supply-chain shocks that exposed just how dependent global industries had become on a handful of manufacturing hubs. But the Commission now believes the challenge has evolved.

The rise of AI has transformed semiconductors from a supply-chain issue into a strategic competition. AI-related components are expected to account for more than 70% of the semiconductor market by 2030, making access to advanced chips one of the defining questions of the next decade.

Chips Act 2.0, therefore, goes beyond manufacturing capacity. Alongside support for research, innovation, and production, it introduces measures designed to stimulate demand for European chips, strengthen chip design capabilities, and encourage closer cooperation between manufacturers and the industries that rely on them.

The proposal also reflects a broader shift in thinking. The first Chips Act focused heavily on supply. The new version recognises that supply and demand are interconnected. If Europe wants a stronger semiconductor industry, it needs European customers willing—and able—to buy European chips.

Building Europe’s cloud. The Cloud and AI Development Act seeks to address what Brussels increasingly sees as one of Europe’s biggest weaknesses: a shortage of computing capacity. As AI systems become larger and more resource-intensive, access to data centres and cloud services is becoming as strategically important as access to energy or transportation infrastructure. Yet Europe remains heavily dependent on foreign providers, particularly US hyperscalers.

That dependence raises economic questions, but it also raises political and legal ones. European governments are increasingly uncomfortable with the idea that critical services, sensitive data, and key parts of the digital economy rely on infrastructure operating under foreign jurisdictions.

The Commission’s answer is ambitious. CADA aims to triple Europe’s data-centre capacity over the next five to seven years while creating a framework to assess the sovereignty of cloud and AI services based on factors such as infrastructure location, supply-chain control, and cybersecurity.

Open source enters the sovereignty debate. For years, open source was discussed primarily in terms of innovation, collaboration, and software development. The Commission is now framing it as a sovereignty issue.

The argument is simple enough. Europe is home to more than three million open-source contributors and hundreds of companies built around open-source technologies. Yet public administrations and businesses continue to spend vast sums on proprietary digital products and services, many of them provided by companies outside Europe. At the same time, European developers often contribute to projects whose economic value is ultimately captured elsewhere.

Brussels sees an opportunity to change that. The strategy aims to support European open-source alternatives in areas such as cloud services, AI, cybersecurity, semiconductors, and digital identity. It also seeks to address some of the long-standing challenges facing the ecosystem, including limited funding, difficulties scaling successful projects, and the often-overlooked problem of maintaining critical software once the initial development work is complete.

No digital sovereignty without energy sovereignty. There is another reality shaping Europe’s digital ambitions: AI requires enormous amounts of electricity. As policymakers race to build more data centres, they are also confronting the question of how those facilities will fit into Europe’s energy system. According to the Commission, data centres already consume enough electricity to power nearly 20 million European households, and demand is expected to more than double by 2030.

The Strategic Roadmap for Digitalisation and AI in Energy seeks to ensure that expanding digital infrastructure does not overwhelm Europe’s energy transition. It promotes closer coordination among energy operators, public authorities, and data centre providers while encouraging the deployment of smart grids, smart meters, and AI-driven energy management systems.

The wider context. The backdrop is a growing concern that Europe has become too dependent on technologies and services controlled elsewhere. The Commission estimates that the EU relies on non-EU providers for more than 80% of its digital products, services, and infrastructure. In cloud computing, three non-European hyperscalers account for more than 70% of the market, while Europe’s own share has steadily declined.

For years, these dependencies were largely seen as the inevitable consequence of globalisation. European businesses gained access to cutting-edge technologies, consumers benefited from competitive services, and policymakers paid little attention to who controlled the underlying infrastructure.

That calculus has changed. The past few years have seen a sharp increase in export controls, investment restrictions, supply-chain disruptions, and efforts by major powers to use their technological advantages as geopolitical leverage. What was once seen as healthy interdependence increasingly looks like—and is—a vulnerability.

The Commission’s response is straightforward. The goal is not to shut down foreign providers. It is the ability to make choices, manage risks, and avoid finding itself locked into dependencies that others can exploit. Europe does not need to produce everything itself, but it wants greater control over the technologies and infrastructure on which its economy depends. Whether that balance can be maintained will be one of the most closely watched aspects of the proposal.

But, can Europe deliver? Europe has become increasingly good at identifying strategic problems. Turning that diagnosis into industrial capacity has proven more difficult. The Commission is asking Europe to build more chips, more cloud infrastructure, more open-source alternatives, and more AI capacity. None of that will happen overnight. The legislative proposals must still pass through the European Parliament and the Council. Significant investment gaps remain. And competitors are not standing still. The harder question is whether Europe can move quickly enough to close the gap.

Weekly #268 Quantum sandcastles: The US races to turn theory into power

22 – 26 June 2026

HIGHLIGHT OF THE WEEK

Quantum sandcastles: The US races to turn theory into power

The USA is accelerating its push into quantum technologies, with President Donald Trump signing two executive orders at the beginning of the week.

The first order (EO 1142) seeks to speed up the migration of government systems to post-quantum cryptography, with key systems expected to transition by 2030 or 2031. The administration argues that current encryption standards could eventually be broken by sufficiently powerful quantum computers, creating long-term risks for government, financial, and critical infrastructure systems.

The second order (EO 1143) directs federal agencies, research institutions, and industry partners to work towards a quantum computer capable of conducting advanced scientific research within the next two years. The order also promotes the deployment of quantum sensors and quantum networks, with agencies tasked to develop five-year plans for commercial, scientific, and space applications. It also seeks to strengthen domestic quantum supply chains by expanding manufacturing capacity, supporting quantum-enabling technologies, encouraging private-sector investment, and reducing dependence on foreign suppliers.

The headline goal of achieving a powerful quantum computer by 2028 is ambitious and uncertain. The more immediate impact may come from the push toward post-quantum cryptography. On the international level, the orders could intensify international discussions on standards, export controls, and security frameworks for quantum technologies.

Person, Book, Comics, Publication, Bag, Face, Head, Outdoors

Why does it matter? The announcement is just the latest development that shows that governments are preparing for a future in which quantum technologies are no longer a distant possibility but an emerging strategic capability. Governments are investing in regulation, standards, security, and migration planning – data shows governments worldwide committing more than €43 billion in dedicated public funding in 2026. Even if the technical timelines prove optimistic, the policy response is already underway, a rare thing in the tech sector.

IN OTHER NEWS LAST WEEK

AI governance

The UN. UN Secretary-General António Guterres has called on AI companies to publicly disclose the environmental impact of their operations, including carbon emissions, water consumption, and land use. Guterres also called on AI firms to commit to powering their facilities entirely from renewable sources such as wind and solar by 2030, though existing clean energy commitments from major tech companies have already been complicated by the rapid pace of AI deployment.

The EU. The European Commission has signed the Pax Silica Declaration on behalf of the EU, joining an international initiative focused on AI security and resilient silicon supply chains. The EU’s signature follows the adoption of the European Technological Sovereignty Package, which includes Chips Act 2.0 and measures to strengthen Europe’s capacity in semiconductors, AI, cloud and open-source technologies.

Estonia. Estonia is moving forward with plans to create official digital identities for AI agents, a move that could make it the first country to establish a formal framework for AI systems acting on behalf of individuals and organisations. Officials argue that digital identities for AI could prevent users from granting excessive access to personal data and services while supporting the growing use of AI across the economy.

Indonesia. A draft presidential regulation outlines plans to integrate AI across key government programmes as part of a broader AI roadmap for 2026–2029. The policy positions AI as a tool for administrative efficiency, service optimisation and economic growth, with projected GDP gains of up to 12% by 2030 if implementation succeeds at scale. The framework proposes using AI in the free-meal scheme to design region-specific menus, forecast demand, monitor kitchen hygiene, detect irregularities and integrate health data for early warning systems. It also extends to public health functions such as disease screening and monitoring. A proposed sovereign AI fund under Danantara would support implementation, alongside fiscal incentives for research and talent development. Government agencies would be required to report AI-related risks, including deepfakes, biometric misuse and intellectual property violations.

Norway. The government has introduced a near-ban on the use of AI in elementary education, restricting AI use for pupils aged 6 to 13 and setting stricter limits across lower school levels. Prime Minister Jonas Gahr Stoere said the measures are intended to prevent children from bypassing essential learning processes in reading, writing and mathematics through early reliance on AI tools. However, pupils in lower secondary school (ages 14–16) may use AI tools only cautiously and under teacher supervision, while upper secondary students (ages 17–19) are expected to develop appropriate AI literacy to prepare for further study and the labour market.

Child online protection

This week, several governments moved toward direct limitations on device and platform use in education and early adolescence.

Czech Prime Minister Andrej Babis stated that the Czech government is preparing a nationwide ban on mobile phones in schools from September 2027. A government-appointed panel in Germany recommended a social media ban for under-13s and tiered protection standards for the 13–16 and 16–18 age groups, seeking to shield young people from online risks without excluding them from digital spaces entirely.

The approval of Spain’s Law for the Protection of Minors in Digital Environments is expected in autumn 2026. Measures in the bill include mandatory parental control systems for mobile devices, rules on loot boxes in video games and on platforms, and requirements for schools to regulate the use of mobile phones and digital devices. The proposal would also introduce criminal law changes covering digital violence. These include penalties restricting aggressors from contacting victims online, offences linked to making pornography indiscriminately available to minors, and criminalisation of sexual or seriously degrading deepfakes. Large audiovisual service providers and major influencers would also be required to provide reporting channels for inappropriate content, inform users about content that may harm minors, use effective age verification systems and separate pornographic or violent content from other material.

In the USA, Meta is pushing lawmakers to include legal protections in the Kids Online Safety Act that would largely shield the company from lawsuits alleging harm to children on platforms such as Instagram. The proposal would shift liability into a federal framework and preempt many state-level claims. Meta argues it is seeking a consistent national standard rather than broad immunity, but critics warn the measure could effectively derail existing and future lawsuits over children’s online safety.

The legal pressure on platforms continues elsewhere. Google’s YouTube has settled a social media addiction lawsuit brought by a 15-year-old in Florida, who alleged that YouTube and other platforms deliberately designed their services to be addictive and harmful to young users. The settlement adds to a growing wave of litigation linking social media design to children’s mental health.

Across the Atlantic, the European Commission is reportedly preparing to escalate its own scrutiny of Meta. According to reports, Brussels is set to issue preliminary findings claiming that Facebook and Instagram are using addictive and exploitative design practices to keep children engaged.

Why does it matter? Children’s online protection has become one of the defining digital policy issues of the past year, and nothing indicates that that momentum is slowing. The simultaneous push for new legislation, tougher enforcement, and private litigation leaves platforms under pressure from all sides. Even as companies seek greater legal certainty, policymakers appear increasingly determined to hold them accountable for how their products are designed and used by children.

Submarine cables on a Thursday

Washington and Brussels took different approaches to the same challenge this Thursday: protecting submarine cables, the largely invisible infrastructure that carries almost all of the world’s internet traffic.

In the USA, the Federal Communications Commission (FCC) voted to tighten oversight of submarine communications cables, introducing a more restrictive licensing regime for submarine line terminal equipment — the systems that connect undersea cables to terrestrial networks. For the first time, operators of this equipment will require FCC licences, a move designed to give regulators greater visibility and control over the most sensitive chokepoints in cable infrastructure.

The framework is also explicitly tilted toward national security screening, making it harder for Chinese-linked companies to supply equipment or participate in high-risk deployments. At the same time, the FCC is streamlining approvals for US technology firms, including major platforms such as Meta and Alphabet, which are expanding their undersea cable portfolios to meet rising global bandwidth demand.

Across the Atlantic, the European Commission is taking a more infrastructure-centric approach. It has announced €8.3 million to establish regional hubs tasked with monitoring and repairing subsea cables, alongside a broader €40 million call for proposals to strengthen repair capacity in the Mediterranean and Atlantic basins.

The new hubs will be split between two regional clusters: the Baltic Sea hub, coordinated by Finland with Denmark, Estonia, Germany, Latvia and Sweden, and a Mediterranean hub led by Italy with participation from Cyprus, Greece and Malta. The Commission’s stated aim is to detect and prevent threats to subsea infrastructure while also reducing repair latency, a known vulnerability in cable systems where outages can take weeks to resolve, depending on ship and parts availability.

The initiative comes amid heightened concern over damage to cables in the Baltic Sea, which EU officials and several member states have linked to hybrid threats and suspected sabotage operations in the context of the war in Ukraine.

Why does it matter? Carrying around 99% of international internet traffic, subsea cables are essential to economic security, national defence, and digital sovereignty. As geopolitical tensions rise, governments are paying as much attention to who builds, operates, and repairs them as to the data flowing through them.

EU targets AWS and Azure cloud computing services under the DMA

The European Commission has informed Amazon and Microsoft of its preliminary view that their cloud computing services, Amazon Web Services and Microsoft Azure, should be designated as gatekeepers under the Digital Markets Act.

The Commission opened market investigations into AWS and Azure in November 2025. It has now been provisionally concluded that both services act as important gateways between businesses and customers in the EU, despite not meeting the DMA’s standard quantitative thresholds.

What’s next? Amazon and Microsoft now have the opportunity to examine the investigation files and respond to the preliminary findings. If the Commission confirms its assessment, AWS and Azure would be designated as gatekeepers, and the companies would have six months to comply with DMA obligations.

Bank of England drops stablecoin holding limits in revised framework

The Bank of England has revised its proposed regulatory approach for sterling-denominated systemic stablecoins, removing planned individual holding limits and introducing a temporary £40 billion issuance guardrail for each systemic stablecoin.

The policy statement and draft Code of Practice set out how the Bank intends to regulate stablecoin issuers that become systemic because their use in payments could pose risks to UK financial stability. The final Code of Practice is expected by the end of 2026.

The Bank said it had changed its approach following industry feedback on two key issues: backing asset composition and holding limits. Earlier proposals would have imposed individual limits of £20,000 per coin for individuals and £10 million for businesses, with possible exemptions.

Instead, the Bank now plans to use a temporary issuance guardrail, initially set at £40 billion per systemic stablecoin. The Bank said the approach would be simpler to implement while still managing credit risks as stablecoins scale.

EDPB adopts common data breach notification template

The European Data Protection Board adopted a common template for GDPR data breach notifications, aiming to simplify compliance and improve consistency across the EU. The template is designed to help organisations and data protection authorities structure breach reports under Article 33 of the GDPR, with predefined fields and guidance intended to reduce administrative burdens.

This is a technical change, but a meaningful one. Data breach notification is one of the practical mechanisms through which privacy law becomes operational. Fragmented national practices can make compliance harder, especially for organisations operating across several EU countries. A shared template may support more predictable enforcement, faster reporting, and better incident comparison across jurisdictions.

View all updates

LAST WEEK IN GENEVA

UNIDIR launches a Centre of Excellence on AI, Peace and Security

UNIDIR has launched a Centre of Excellence on AI, Peace and Security in Geneva to address gaps in the governance of AI in the peace and security domain.

While AI has become a priority across international policy forums, efforts remain fragmented across regions and institutions, precisely the gap the centre aims to address. It is designed as a permanent platform to consolidate knowledge and expertise, enable sustained multistakeholder dialogue, and support the translation of high-level principles into practical governance and operational approaches.

The approach combines three strands: generating evidence-based knowledge on emerging risks, facilitating inclusive dialogue among governments, industry, and expert communities, and supporting action-oriented pathways that move policy discussions toward implementation.

Rather than duplicating existing initiatives, the centre positions itself as a complementary coordination space.

Global Conference on AI, Security and Ethics 2026

The UNIDIR Global Conference on AI, Security and Ethics 2026 (#AISE26) was held on 18–19 June in Geneva and online, bringing together diplomats, policymakers, researchers, industry representatives, civil society, and military experts to examine the implications of AI for international peace and security.

Across two days, discussions were structured around technology foundations, applications and integration, stakeholder perspectives, and forward-looking governance approaches. Sessions examined issues such as algorithmic bias, dual-use AI systems, adversarial testing, and the integration of AI into defence and security infrastructures. Particular attention was given to counter-AI risks, infrastructure dependencies, and the growing role of autonomous or agentic systems.

The second day shifted toward governance dynamics, including investor responsibility, public–private cooperation, regional AI strategies—especially in Africa and the Global South—and compliance challenges linked to emerging uses such as deepfakes in conflict contexts. Speakers also highlighted tensions between innovation, accountability, and international humanitarian law.

The conference featured fireside chats, technical demonstrations, lightning talks, and panel discussions, aiming to bridge technical and policy communities.

The 62nd session of the HRC continues

The 62nd session of the UN Human Rights Council (HRC) is ongoing, with this week’s programme featuring several discussions on digital governance.

On Monday, discussions included ‘From Rabat to Glion – Safeguarding Freedom of Expression and Freedom of Religion or Belief in the Digital Age’, co-organised by the Permanent Missions of Albania, Switzerland, Morocco, OHCHR, and Article 19, as well as ‘From Rights to Architecture: Women, Girls, and the Governance of Artificial Intelligence’, supported by a broad coalition including Japan, France, Estonia, Armenia, Costa Rica, Mexico, and the Republic of Korea.

On Wednesday, attention shifted to child protection in digital environments with an event on ‘Protecting Children from Emerging Forms of Harm in the Digital Era’, led by Egypt and supported by multiple states and organisations. The Special Rapporteur’s report on the chilling effects of digital surveillance on assembly and association rights (A/HRC/62/45) was also presented at a side event organised by the Permanent Mission of the Czech Republic and the OHCHR Special Rapporteur on the promotion and protection of the right to freedom of opinion and expression. The week also included a Thursday session on child protection in digital spaces, organised by the United Arab Emirates and the Permanent Committee for Human Rights.

The HRC is in session until 7 July.

First global rules for fully driverless cars adopted by UNECE

The UNECE World Forum for Harmonization of Vehicle Regulations (WP.29) has adopted the world’s first global regulatory framework enabling fully autonomous driving systems (ADS).

The rules establish uniform safety requirements based on a ‘safety-case’ approach, requiring manufacturers to demonstrate through testing, simulation, and real-world validation that autonomous systems perform at least as safely as competent human drivers.

The framework introduces mandatory Safety Management Systems, strict testing standards, continuous in-service monitoring, and data recording requirements, while also updating around 90 existing UN vehicle regulations to accommodate automation. Backed by major automotive markets including the EU, the USA, China, Japan, the UK, and Canada, the agreement aims to prevent fragmented national rules and provide a coordinated pathway for global deployment of autonomous vehicles.

READING CORNER

Geneva at the centre of AI governance: Where technology, diplomacy, and humanity converge

In an interview with Diplo’s Executive Director Jovan Kurbalija, the city of Geneva is presented as a unique meeting point where technology, diplomacy, history, and humanity converge to shape global discussions on the future of AI.

digital map with illuminated grid lines representing african continent 1

Africa’s digital diplomacy in the AI era: Building a common voice for global digital governance

Experts participating in a Diplo webinar on cyber diplomacy in Africa highlighted the importance of stronger regional coordination, greater participation in global digital governance, and investment in local digital capacity.

UN report outlines AI standards for Digital Public Goods

Findings suggest AI systems should be assessed through public value frameworks prioritising equity, governance and societal impact alongside openness.

OECD publishes AI literacy framework for schools

The AI literacy framework aims to support learning outcomes for primary and secondary students worldwide.

OECD report highlights AI’s growing role in workforce training

A new OECD report highlights opportunities and risks of AI in VET systems.

DON’T MISS

Follow the road to the 2027 Geneva AI Summit, a global journey shaping AI’s future. Bookmark the page and register for updates to stay informed and involved every step of the way to Geneva 2027.

Weekly #267 Who gets to pull the plug on frontier AI?

12 June – 19 June 2026

HIGHLIGHT OF THE WEEK

Who gets to pull the plug on frontier AI?

Access to powerful AI systems is no longer just a product feature, a subscription tier, or a matter of corporate safety policy. It is becoming a governance decision. When an AI model is capable enough to support software engineering, scientific research, cybersecurity work, critical infrastructure protection, and potentially harmful misuse, the question is no longer only how well the model performs. The question becomes who should be allowed to use it, under what conditions, and who has the authority to switch it off.

That question moved from theory to practice over the past week, after Anthropic disabled access to Claude Fable 5 and Claude Mythos 5 in response to a US government export-control directive. The timing made the case especially striking. Only days earlier, Anthropic had launched Fable 5 as its most capable generally available model to date, alongside Mythos 5, a more powerful version reserved for selected cyber defence and critical infrastructure partners.

At first, the launch looked like a model governance experiment. Fable 5 was made broadly available, but with additional safety classifiers designed to detect potential misuse, including attempts to bypass safeguards. Anthropic also introduced a fallback system that routed some high-risk sessions to a less capable model. Mythos 5, meanwhile, used the same underlying model, but with some safeguards lifted for trusted users working in cyber defence and critical software infrastructure.

In other words, Anthropic was trying to answer one of the hardest questions in frontier AI deployment: how can companies give users access to more capable systems while reducing the risk that the same systems are misused in sensitive domains? Its answer was tiered access, safety filters, model routing, testing, and trusted partnerships.

Then came the cutoff.

What followed was more than a technical compliance problem. According to Anthropic, the US directive required the company to suspend access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The company said the practical result was that it had to remove access to both models for all customers to ensure compliance, while other Anthropic models remained available.

Why is this happening? The immediate reason centres on concerns that Fable 5 could be jailbroken. Anthropic said it understood that the government had become aware of a method for bypassing the model’s safeguards, but also said the directive did not provide specific details of the national security concern. The company argued that the technique it reviewed appeared to identify a small number of previously known minor vulnerabilities, which could also be found using other publicly available models.

That disagreement matters because it points to a much larger governance problem. Governments clearly have a legitimate interest in preventing the misuse of powerful AI systems, especially when they may affect cybersecurity, biological research, defence, or critical infrastructure. But companies, users, researchers, allies, and civil society also have an interest in knowing when a deployed model can be recalled, what evidence is required, and what transparency or due-process safeguards apply.

If the standard for suspending access is too weak, frontier AI deployment becomes unpredictable. If the standard is too high, genuinely dangerous systems may remain available for too long. The difficult part is building a system that can respond quickly in real emergencies without allowing access to advanced AI to become an unclear political decision.

The episode also gives new force to the debate on AI sovereignty. Governments outside the United States have spent the past year asking whether they can rely on AI systems controlled by a small number of foreign companies. The Fable 5 and Mythos 5 restrictions made that question less abstract. If foreign users, trusted partners, and even allied institutions can lose access overnight due to a US national security decision, then reliance on frontier AI providers becomes a strategic vulnerability.

That does not mean every country can build its own frontier models. AI sovereignty is easier to demand than to deliver. Training and deploying advanced systems requires chips, data centres, energy, skilled researchers, secure cloud infrastructure, capital, and access to global supply chains. For many governments, the real choice may not be between full dependence and full sovereignty, but between managed dependence, trusted access, regional capacity, and strategic resilience.

For the United States, there is also a trade-off. Broad restrictions may reduce certain security risks, but they can also erode trust with allies, disrupt research cooperation, and encourage other countries to accelerate the development of domestic alternatives. AI leadership is not only about having the most capable models. It also depends on whether partners believe that access to those systems is predictable, lawful, and governed by clear rules.

The Fable 5 saga shows that frontier AI is becoming more like strategic infrastructure than ordinary software. Access decisions can affect cybersecurity teams, research institutions, public agencies, private companies, and international partnerships. Safety cannot rely solely on company policy, but government intervention also cannot rest on vague national security claims.

The bigger question is whether societies can design rules for frontier AI access that are fast enough to address real risks, transparent enough to build public trust, and stable enough to support international cooperation. If powerful models can be launched one week and pulled back the next, the politics of AI will increasingly revolve not only around what these systems can do, but around who gets to decide who can use them.

IN OTHER NEWS LAST WEEK

AI governance and digital security

G7. Leaders of the G7 nations gathered in Évian-les-Bains, France, for the 2026 summit, with critical mineral supply chains, online child safety and international solidarity among the key issues on the agenda. France placed balanced economic growth and international cooperation at the centre of this year’s discussions, while leaders also addressed geopolitical tensions, support for Ukraine, regional security and the Middle East. The digital policy dimension was visible through debates on resources needed for semiconductors and digital infrastructure, as well as the growing connection between economic security, technology governance and geopolitical stability.

The EU. The European Commission welcomed the new G7 Cybersecurity Working Group Declaration, adopted under France’s G7 Presidency. The declaration calls for stronger international cooperation on cybersecurity challenges linked to quantum computing, AI, telecommunications infrastructure and the protection of small and medium-sized enterprises. It treats the transition to post-quantum cryptography as an urgent priority, while also recognising that AI can both strengthen and threaten cybersecurity through AI-enabled attacks, model manipulation, data breaches and software vulnerabilities.

G7. Digital and technology ministers adopted the group’s first shared set of principles for protecting children and young people online. The agreement addresses risks linked to harmful content, exploitation and AI chatbots, while calling for stronger digital literacy, safety measures built into digital services from the start, and effective age assurance. Ministers also called for better access to data and research on how digital services affect children’s well-being, alongside closer cooperation among platforms, researchers, families and governments.

The EU. The European Commission also welcomed the G7’s adoption of online child protection principles, noting that they reflect approaches already present in EU initiatives such as the Digital Services Act, the Better Internet for Kids Strategy and the AI Act. The framework promotes safety-by-design measures, privacy-conscious age assurance tools, stronger protections against harmful and illegal content, parental controls and digital literacy. It also links child protection with privacy, fundamental rights and access to digital opportunities, showing how online safety policy is moving beyond content removal alone.

UNIDIR launches AI peace and security platform in Geneva

The UN Institute for Disarmament Research, together with Switzerland and Pakistan, hosted a pre-launch briefing for its Centre of Excellence on AI, Peace and Security at the Palais des Nations in Geneva. UNIDIR said the centre is being established at a critical moment for global AI governance, as AI increasingly reshapes international peace and security dynamics. The centre is intended to serve as a permanent platform for consolidating knowledge, connecting stakeholders and maintaining continuity between multilateral processes and global discussions on AI and international security.

The platform will support more coherent international AI governance while promoting inclusive global engagement. It will provide practical, evidence-based policy guidance, resources and capacity-building support for actors working on AI in peace and security contexts. The initiative also points to Geneva’s growing role as a hub where AI governance, security policy, disarmament debates and multistakeholder dialogue increasingly overlap.

Children’s online safety under UK scrutiny

The UK Government Office for Science highlighted the need for evidence-led policy on children’s online lives, warning that digital technologies bring both benefits and risks while long-term evidence remains limited. The article noted that 97% of UK teenagers aged 13 to 15 now own a mobile phone, while almost one-fifth of children aged three to five also own one. Children use digital platforms to maintain friendships, access communities, and find support, while AI tools are increasingly used for learning, schoolwork and, in some cases, well-being and emotional regulation.

However, the government adviser warned that children face risks including harmful content, cyberbullying, privacy breaches, false or misleading information, unlimited scrolling, personalised algorithms and other engagement-maximising design features. The article also cautioned that the use of AI should not prevent children from developing skills such as written expression and critical thinking. The debate shows why child online safety is increasingly being framed not only as a platform moderation problem, but as a question of design, evidence, education and long-term development.

IWF warns of AI-generated child sexual abuse risks

The Internet Watch Foundation welcomed Pope Leo XIV’s reflections on AI, arguing that AI systems must be developed with stronger safeguards to protect children from abuse. The organisation warned that AI is being used to generate highly realistic child sexual abuse images and videos at scale, saying the number of AI-generated child sexual abuse videos it identified in 2025 increased by more than 260%, with nearly two-thirds falling into the most severe category of abuse.

The IWF also raised concerns about AI-nudification tools, which can generate realistic sexualised images of children and other individuals. Following the Child Dignity in the Artificial Intelligence Era conference in Rome, the organisation joined more than 100 organisations and individuals in supporting calls for a global ban on such tools. It also called for stronger regulation, global alignment and enforceable safety-by-design standards to prevent the creation and spread of AI-generated child sexual abuse images and videos.

The EU extends cyber crisis support to Ukraine

Ukraine can now activate emergency EU cyber support during significant or large-scale cybersecurity incidents after the Council of the EU approved its inclusion in the EU Cybersecurity Reserve. The reserve, managed by the European Union Agency for Cybersecurity, provides incident response services from trusted private-sector providers to help contain and mitigate major cyber incidents.

The European Commission said the decision reflects closer EU-Ukraine cooperation and forms part of wider efforts to strengthen preparedness, rapid response and shared expertise against evolving cyber threats. The move also aligns with the EU’s strategic digital partnership agenda and follows Moldova’s inclusion in the Cybersecurity Reserve in 2024 under the Cyber Solidarity Act. Ukraine’s inclusion shows how the EU is using cyber crisis mechanisms not only for internal resilience but also for strategic partnerships in a conflict-affected digital environment.

Latin America looks to Europe’s AI rulebook

Countries across Latin America are increasingly drawing on the EU’s risk-based approach as they develop AI governance frameworks, according to an International Bar Association analysis. Chile, Peru, Brazil and El Salvador have introduced or advanced AI-related legislation in recent years, with several initiatives focusing on risk classification, fundamental rights and accountability. Peru implemented an AI law in 2023 and updated it in 2025, while Brazil has considered similar draft legislation, and El Salvador enacted an AI law in 2025.

Chile has become one of the region’s most active AI policy actors, helping launch Latam-GPT, an open-source language model trained on Latin American data, while considering an AI bill that would classify systems according to risk. The proposal includes stricter obligations for high-risk uses, regulatory sandboxes and measures intended to support innovation and smaller businesses. However, legal experts warned that EU-style requirements on risk governance, technical documentation, testing and incident reporting could strain regulators and smaller companies with limited resources.

Swiss parliament weighs AI apps in media copyright bill

Swiss lawmakers want the government to examine whether AI applications should be covered by a media copyright bill that requires online services to compensate publishers for displaying extracts of newspaper articles. The Swiss Senate unanimously referred the bill back to the federal government, while the House of Representatives had already approved the request in March by 157 votes to 29, with two abstentions.

Why does it matter? The debate shows how AI-powered search tools, chatbots and assistants are being pulled into older disputes over platform power, copyright and publisher compensation. As AI services become gateways to information, governments are asking whether existing media copyright rules remain fit for a digital environment shaped not only by search engines and social platforms but also by AI applications.

MIT researchers target AI data centre cooling

A startup founded by MIT researchers has developed a nuclear-inspired cooling system to improve data centre energy efficiency while reducing water consumption. The technology targets one of the fastest-growing sources of electricity demand, as the rapid expansion of AI infrastructure drives increased computing requirements.

Ferveret’s system uses a specialised liquid to immerse servers, replacing traditional air-based cooling methods that account for a significant share of data centre energy consumption. Its Adaptive Phase Cooling approach improves heat transfer through controlled bubble formation, increasing efficiency while reducing reliance on water-intensive cooling systems. The company reports computational efficiency gains of up to 15% compared with existing liquid cooling technologies and says its modular design could help data centres operate more effectively in regions with limited water resources.

View all updates

LAST WEEK IN GENEVA

Geneva placed AI, security and ethics at the centre of international security diplomacy this week. From 15 to 17 June, UNODA convened informal exchanges on AI in the military domain and its implications for international peace and security, following the UN General Assembly’s request for states and stakeholders to discuss observations from the Secretary-General’s report, opportunities and challenges, emerging normative proposals and possible next steps.

The exchanges brought together states, observer states, international and regional organisations, the ICRC, academia, civil society, the scientific and technical community and industry. Discussions and side events addressed the responsible development and use of AI in military decision-making, trustworthy AI in the military domain, human rights across the life cycle of military AI, assurance and accountability, the role of private actors, weaponisation of data-intensive technologies, and the nuclear dimension of AI.

The Geneva week then continued with UNIDIR’s Global Conference on AI, Security and Ethics 2026, held on 18 and 19 June in Geneva and online. The conference brought diplomats, policymakers, military representatives, researchers, industry, civil society and international organisations together to examine how AI is reshaping national, regional and global security environments. The programme focused on themes ranging from algorithmic bias, dual-use technologies, agentic AI, testing and assurance, and AI in cyber defence to counter-AI capabilities, infrastructure dependencies, public-private responsibility chains and human-centred approaches to defence AI.

Together, the UNODA exchanges and UNIDIR conference showed how Geneva’s AI diplomacy is moving beyond broad concern about emerging technologies. The agenda is increasingly about how to translate international law, responsible AI principles, military practice, technical testing and multistakeholder expertise into workable governance for high-stakes systems. As AI becomes embedded in defence, cybersecurity and strategic infrastructure, Geneva is becoming one of the key spaces where the security implications of AI are debated not only as future risks, but as present governance challenges.

LOOKING AHEAD

The Developments in Africa cyber diplomacy: Continental, regional, and national initiatives webinar will take place online on 24 June, organised by Diplo. The event will examine the evolving landscape of cyber diplomacy across Africa, focusing on continental, regional and national initiatives. Discussions will cover key players, stakeholders and frameworks shaping cyber diplomacy on the continent, including African Union initiatives and the African Digital Compact, as well as national strategies for addressing cyber threats and strengthening digital cooperation. The webinar aims to bring together policymakers, cybersecurity experts, researchers, civil society and private sector stakeholders to exchange lessons, identify best practices and support secure and inclusive digital transformation across the continent.

UN Open Source Week 2026 will take place from 22 to 26 June at UN Headquarters in New York, co-hosted by the UN Office for Digital and Emerging Technologies and the UN Office of Information and Communications Technology. The event is framed as a global convening on open-source technologies and digital cooperation, with a programme covering a Community Hackathon, open source and AI, Digital Public Infrastructure Day, Open Source Programme Offices for public-interest objectives and community-led events. The week will explore how open source can support the SDGs, the Global Digital Compact, digital public infrastructure, responsible AI development and practical collaboration among UN entities, member states, technical communities, civil society and the private sector.

READING CORNER

united kingdom flag wrinkled dark background 3d render

The future of agentic AI: A cross-regulatory perspective from the UK

Agentic AI systems that can plan, decide and act autonomously across platforms simultaneously trigger obligations under data protection, competition, financial services and online safety law, exposing gaps that sector-based regulation…

Fable 5 and the gatekeeping of human reasoning

Anthropic’s Claude Fable 5 went public, then vanished 72 hours later. Inside the sudden U.S. government shutdown and what it means for the future of AI power.

DON’T MISS

Follow the road to the 2027 Geneva AI Summit, a global journey shaping AI’s future. Bookmark the page and register for updates to stay informed and involved every step of the way to Geneva 2027.

Weekly #266 The AI IPO rush

5 June – 12 June 2026

HIGHLIGHT OF THE WEEK

The AI IPO rush

When investors’ demand for a company’s shares far exceeds the number of available shares, they are showing strong confidence that the company could become a major industrial power. At the same time, such intense demand can be a warning sign that expectations are running ahead of proven returns. Surges like this usually come from a mix of real technological potential and market euphoria.

In the past week, SpaceX sought to raise around $75 billion through its IPO (Initial Public Offering), while investor demand approached four times that amount, or more than $250 billion in orders. That means the IPO is, technically speaking, oversubscribed.

OpenAI and Anthropic have also joined the race towards public markets, showing how AI has moved from software promise to financial and geopolitical infrastructure. Governance debates have long focused on ethics, safety, bias, transparency, and regulation, but the IPO wave adds another layer: finance. Once AI companies enter public markets, their development is shaped not only by engineers, founders, researchers, and regulators, but also by quarterly expectations, institutional investors, pension funds, retail shareholders, and market narratives.

What does this mean? Following the logical thread above, AI is becoming an omnipresent metasystem of infrastructure. Developing and making available AI systems requires chips, energy, data centres and data flows, cooling, cloud platforms, secure connectivity, payment systems, and more. In this sense, the race to list OpenAI, Anthropic, and SpaceX is also a race to finance the foundations of the global AI economy. SpaceX is particularly important because strong IPO demand shows that investors see it as more than a rocket company. They are buying into satellites, connectivity, defence relevance, orbital infrastructure, and the role of space-based systems in the wider digital stack. Their listings test whether markets can treat frontier AI as both a technological breakthrough and a sustainable business model. Yet AI systems remain expensive to train and run, and public markets are only one source of funding.

The buildout is also moving into bond markets, with Morgan Stanley expecting global AI-related debt issuance to more than double to nearly $570 billion in 2026 as companies fund chips, data centres, cloud capacity, and related infrastructure. The governance question, therefore, goes beyond IPO valuation, since AI finance is spreading across equity, debt, private credit, and infrastructure investment instruments.

Book, Comics, Publication, People, Person, Face, Head, Crowd, Audience

Why is this happening?

First, investors believe that AI is a new general-purpose technology, like electricity, the internet, or mobile telephones, that will not be limited to just one industry sector, but will be a layer across all industries, like healthcare, finance, education, defence, logistics, media, public administration, and more. That is why money is not flowing only into ‘AI applications’, but also into other areas such as chips, data centres, cloud infrastructure, energy, fibre optics, satellites, and network infrastructure.

Second, there is a first-mover advantage effect. Major companies are spending huge sums because they fear that, if they fall behind, they will lose access to infrastructure, data, models, talent, and users. Even if not every investment is individually rational, no major player wants to be the one that ‘saved money’ while others were building the platform of the future.

Third, investors are buying into future monopoly positions, not just current profits. If a company becomes key infrastructure for AI, like Nvidia in chips, cloud platforms in computing, or SpaceX/Starlink in connectivity and orbital infrastructure, markets value it as a potential gateway through which enormous future revenues may flow. That is why demand for certain IPOs can be so high. Investors are not only buying shares, but also early access to a possible dominant layer of the future economy.

Does the AI IPO growth make sense? Partly yes, but not at any price. It makes sense because AI is already creating measurable economic value in areas such as software development, analysis, and customer support. But it does not make sense when every company with an AI label is valued as if it were a future monopoly.

The best analogy is the internet around 2000. The internet did change the world. But many dot-com valuations were unrealistic and collapsed. Something similar may happen with AI. Few winners may justify enormous investment, while many weaker projects, inflated valuations, and companies without clear revenue models may fail.

There’s also a geopolitical dimension. China’s recently reported state-backed AI data-centre buildout shows that governments also treat AI as strategic infrastructure. On the other hand, US scrutiny of Chinese technology firms through Pentagon-linked lists and debates over investment restrictions shows that AI capital is increasingly filtered through national security concerns.

Wrapping up, public listings can bring benefits like greater transparency, broader ownership, employee liquidity, and stricter disclosure rules. However, they can also deepen the concentration of power if a small number of firms absorb most capital, talent, infrastructure, and political attention.

The question is: Can societies finance innovation without surrendering oversight, scale infrastructure without entrenching dependency, and build powerful AI systems without losing sight of the people and communities they are meant to serve?

IN OTHER NEWS LAST WEEK

AI governance

The USA. The US government published a new AI Executive Order 14409 on innovation and security, Promoting Advanced Artificial Intelligence Innovation and Security, in the Federal Register on 5 June. The order frames advanced AI as both an innovation priority and a security issue, signalling a governance approach focused on maintaining US leadership while addressing risks linked to powerful AI systems.

Brazil. Brazil’s telecommunications regulator, Anatel, adopted a new AI governance policy on 5 June, setting principles for the ethical, secure, and transparent use of AI in its regulatory and administrative work. Developed by the agency’s IA.lab research group, the framework includes human oversight, transparency, data security, protection of fundamental rights, risk monitoring, and a permanent forum to assessing AI deployment. It also links AI governance with technological sovereignty by encouraging the development of Brazil-built AI solutions, the use of local data, and the development of Portuguese-language applications.

The EU. The EU tech chief Henna Virkkunen defended Europe’s approach of combining AI competitiveness with strong regulation. She said there was no immediate need to revise the EU AI Act for AI agents, while also linking AI policy to Europe’s wider ambitions in semiconductors, quantum, cybersecurity, industrial uptake, and international digital partnerships.

ILO. At the 114th International Labour Conference in Geneva, AI governance was addressed in the Director-General’s report, A moment of choice: Harnessing AI for decent work. The report frames AI as a labour-governance issue, focusing on jobs, skills, rights, social dialogue, inequality, and institutional capacity, rather than treating AI only as a productivity tool.

Ireland. Ireland’s Dáil debated and passed a bill to protect people’s voices, images, likenesses, and digital identities from AI-enabled misuse. The proposal targets deepfakes and other synthetic impersonation harms, including fraud, deception, and unauthorised commercial use of a person’s identity, adding to the wider trend of AI-specific rules for synthetic media.

Germany. Germany announced plans to establish an AI safety institute to assess the performance and risks of advanced AI models. The decision followed a National Security Council discussion on the cybersecurity implications of advanced AI, placing Germany among the growing number of countries building dedicated technical capacity to evaluate AI risks.

The USA. The White House issued a National Security Presidential Memorandum on AI during the week, treating AI as a transformative national security technology. The memorandum links AI adoption to defence, strategic competition, military effectiveness, and the need to manage security risks, showing how AI governance is increasingly being embedded in national security policy.

New York. A New York law requiring advertisements that use AI-generated ‘synthetic performers’ to clearly disclose them took effect in June. The measure is part of a growing state-level approach to AI transparency, especially in advertising, media, and the creative industries, where regulators are addressing synthetic likenesses, consumer awareness, and protection of performers.

China. Reports that China is preparing a large state-backed AI data-centre buildout showed how AI governance is increasingly tied to infrastructure, energy, chips, and strategic autonomy. The development places AI not only in the domain of model regulation, but also in industrial policy, compute capacity, and geopolitical competition over the foundations of the AI economy.

Apple delays Siri AI in Europe amid DMA clash

Apple’s decision to delay the launch of its advanced AI-powered Siri in the EU has become one of the week’s most visible platform-governance disputes. The company blamed the EU’s Digital Markets Act, arguing that interoperability obligations could weaken privacy and security if rival AI assistants are given deeper access to Apple systems. The European Commission rejected the claim that the DMA prevents Apple from launching new services, noting that the company has not yet offered a compliant solution.

The case captures a central tension in digital regulation: how to open gatekeeper ecosystems to competition without undermining user protection. It also shows that AI assistants are becoming part of the next platform-control debate. If Siri, Gemini, ChatGPT, Claude, and other assistants become gateways to search, apps, payments, smart devices, and personal data, then rules on interoperability will shape not only competition policy but also the everyday architecture of AI-mediated services.

The EU publishes final code for labelling AI-generated content

The European Commission published the final Code of Practice on marking and labelling AI-generated content, offering practical guidance ahead of the EU AI Act’s transparency obligations. The code covers machine-readable marking of AI-generated or manipulated text, image, audio, and video, as well as labelling of deepfakes and AI-generated public-interest content.

Why does it matter? The AI Act’s transparency rules are now moving from legal text to operational practice. Labelling standards could shape how platforms, AI developers, media organisations, public institutions, and advertisers handle synthetic content, especially where undisclosed AI use may affect elections, journalism, public debate, scams, or personal reputation. The challenge will be to make labels useful and reliable without creating a false sense of security: bad actors may ignore labelling rules, while legitimate actors carry the compliance burden.

EDPB adopts common data breach notification template

The EU orders Meta to restore access for AI assistants

The European Commission ordered Meta to restore access to WhatsApp for rival general-purpose AI assistants while an antitrust investigation continues. The interim measures require Meta to reinstate access to the WhatsApp Business Solution under the previous terms, after the Commission found that Meta’s restrictions could harm competition in the emerging market for AI assistants.

Why does it matter? WhatsApp is not only a messaging app. In many markets, it is a default channel for businesses, public services, customer support, and daily communication. If dominant platforms can decide which AI assistants can reach users through such channels, they may shape the AI market before it fully matures. The case links competition policy with AI governance: access to messaging infrastructure could become as important as access to app stores, search engines, or cloud capacity.

View all updates

LAST WEEK IN GENEVA

The 114th International Labour Conference in Geneva put digital labour governance firmly on the agenda. Running from 1 to 12 June, the conference brought together governments, employers and workers from the ILO’s 187 member states, with platform work among the key items under negotiation. Delegates are considering a draft Convention supplemented by a Recommendation on decent work in the platform economy, marking a possible step towards new international labour standards for app-based and platform-mediated work.

AI also featured prominently through the ILO Director-General’s report, A moment of choice: Harnessing AI for decent work. The report frames AI not only as a productivity tool, but as a governance challenge for jobs, skills, inequality, institutions and social dialogue. Together, the platform economy negotiations and AI report show how Geneva’s labour diplomacy is increasingly engaging with the rules, rights and safeguards needed for a digitalised world of work.

LOOKING AHEAD

The 2026 G7 Leaders’ Summit will take place from 15 to 17 June in Évian-les-Bains, France, bringing together the leaders of Canada, France, Germany, Italy, Japan, the UK, and the USA, as well as the EU representatives. Under the French G7 Presidency, discussions will focus on structural imbalances in the global economy, international cooperation, and dialogue among major economic powers, building on preparatory ministerial and technical meetings across foreign affairs, finance, trade, environment, digital policy, development, and home affairs.

The 62nd regular session of the UN Human Rights Council will convene from 15 June to 16 July 2026 at the Palais des Nations in Geneva, providing a key forum for states and stakeholders to address global human rights issues. Digital governance will feature across several reports, including on the use of digital tools for stakeholder engagement in line with the UN Guiding Principles on Business and Human Rights, state and business responsibilities across the lifecycle of emerging digital technologies, the impact of digital surveillance on assembly and association rights, women’s and girls’ rights in relation to AI and digital technologies, and cyberbullying against older persons.

The Global Conference on AI, Security and Ethics 2026, organised by UNIDIR, will convene on 18–19 June at the Palais des Nations in Geneva and online, bringing together diplomats, policymakers, military representatives, researchers, industry, civil society, and international organisations to examine AI’s implications for international peace and security. The programme will focus on AI and security governance, from algorithmic bias, dual-use technologies, agentic AI, and testing of defence-related AI systems to counter-AI capabilities, infrastructure dependencies, resilience, human-centred operational considerations, public–private cooperation, accountability, and links with wider disarmament and security regimes.

READING CORNER

BLOG featured image 2026 59 Interference in the internal affairs of a state under the VCDR

Interference in the internal affairs of a state under the VCDR – Diplo

Article 41 of the VCDR prohibits interference in a state’s internal affairs, yet international law has widened the scope for lawful diplomatic interference. Alan Franklin explores the growing tension between sovereignty and diplomatic practice.

The geography of AI optimism – Diplo

Discover why Asia-Pacific embraces AI as progress while the West views it as a risk. Inside the global AI divide and what it means for future governance.

The EU’s Tech Sovereignty Package and the future of European digital power

Four interconnected measures, including two new laws, could fundamentally reshape how Europe invests in, deploys, and governs the semiconductors, cloud infrastructure and software its economy depends on.

DON’T MISS

Follow the road to the 2027 Geneva AI Summit, a global journey shaping AI’s future. Bookmark the page and register for updates to stay informed and involved every step of the way to Geneva 2027.

Lettre d’information du Digital Watch – Numéro 110 – Mensuelle Mai 2026

Le mois dernier en matière de gouvernance de l’IA

Gouvernance mondiale de l’IA

Le Vatican a mis en place une Commission interdicastérielle sur l’intelligence artificielle, approuvée par le pape Léon XIV, afin de coordonner les travaux sur les implications des technologies d’IA en pleine évolution.

À l’ONU, les préparatifs s’accélèrent en vue du premier Dialogue mondial sur la gouvernance de l’IA, qui se tiendra à Genève les 6 et 7 juillet 2026. L’organisation a invité les États membres et les parties prenantes à désigner des coprésidents pour les discussions thématiques et à participer au premier Dialogue mondial sur la gouvernance de l’IA à Genève, prévu les 6 et 7 juillet 2026. Les discussions s’articuleront autour de quatre thèmes, chacun étant coprésidé par un État membre et un représentant des parties prenantes, dans le but de favoriser les échanges multipartites sur les expériences, les meilleures pratiques et la coopération en matière de politiques. Les gouvernements sont invités à désigner des représentants de haut niveau, tandis que les parties prenantes sont encouragées à désigner des experts chevronnés compétents sur le thème sélectionné.Par ailleurs, le Bureau des Nations unies pour les technologies numériques et émergentes a lancé le Laboratoire de gouvernance de l’IA pour l’humanité à Valence afin de renforcer la coopération internationale en matière de gouvernance de l’IA. Ce Laboratoire s’attachera à améliorer l’interopérabilité entre les cadres de gouvernance nationaux et régionaux et à soutenir la mise en œuvre pratique dans toutes les régions et tous les secteurs. Ses travaux porteront notamment sur la mobilisation de réseaux, l’analyse comparative des politiques et le développement d’outils de coopération pour la gouvernance de l’IA.

Partenariats

L’Australie et le Japon ont renforcé leur coopération en matière de sécurité économique et de technologies critiques.

L’UE et le Japon ont fait progresser leurs travaux conjoints sur la gouvernance de l’IA et les flux transfrontaliers de données. L’Inde et la France ont discuté d’un renforcement de leur coopération dans les domaines de l’espace, de l’IA, des mathématiques appliquées et des technologies de pointe. Le Canada et l’Espagne ont signé un protocole d’accord visant à consolider leur collaboration en matière de développement et d’adoption de l’IA, ainsi que d’innovation numérique. La Corée du Sud et les Pays-Bas ont discuté d’une coopération dans les domaines des semi-conducteurs et de l’IA. La Corée du Sud et les Émirats arabes unis ont élargi leur coopération en matière d’infrastructures d’IA et de semi-conducteurs par le biais d’un nouveau forum bilatéral sur l’investissement axé sur les écosystèmes d’IA, les centres de données et les technologies de pointe. Les États-Unis et la Suède ont signé un nouveau protocole d’accord axé sur la coopération en matière de technologies stratégiques, de recherche et d’innovation industrielle. La Russie et la Chine ont convenu de coordonner leurs positions sur les questions scientifiques et politiques liées à l’IA au sein des organisations internationales. Même les rivaux géopolitiques semblent rouvrir prudemment le dialogue sur la gouvernance de l’IA. Les États-Unis et la Chine envisagent d’entamer des discussions officielles sur l’IA, reconnaissant ainsi qu’un certain degré de coordination pourrait devenir inévitable à mesure que les systèmes de pointe gagnent en capacité et en importance à l’échelle mondiale.

Stratégies nationales en matière d’IA

En mai, les gouvernements ont continué à affiner leurs programmes nationaux en matière d’IA.

La Maison Blanche a publié un décret visant à renforcer la cybersécurité américaine grâce à une coopération plus étroite avec le secteur de l’IA, tout en conservant une approche modérée de sa réglementation afin de préserver le leadership technologique américain. Ce décret enjoint Ce décret charge les agences fédérales d’accélérer la protection des réseaux gouvernementaux, d’élargir l’accès aux outils de cybersécurité basés sur l’IA et d’améliorer le soutien apporté aux opérateurs d’infrastructures critiques, notamment les hôpitaux, les banques et les services publics. Le décret établit également un centre d’échange volontaire sur la cybersécurité en matière d’IA afin de coordonner la détection des vulnérabilités et les efforts de correction au sein du gouvernement et de l’industrie. En outre, les autorités fédérales mettront au point un processus classifié pour évaluer les systèmes d’IA avancés et identifier les « modèles de pointe couverts » dotés de capacités cybernétiques significatives. Les développeurs seront encouragés, sans y être tenus, à fournir au gouvernement un accès anticipé à ces modèles avant leur diffusion à plus grande échelle.

La Papouasie-Nouvelle-Guinée a défini une approche nationale de l’IA axée sur la souveraineté des données, des infrastructures publiques fiables et une nouvelle législation, s’appuyant sur quatre éléments du cadre : le renforcement des fondements numériques existants tels que SevisPass et SevisDEx, la création d’un registre national de l’IA, l’adoption d’une gouvernance souveraine des données, et l’introduction de nouvelles lois, notamment une loi nationale sur l’intelligence artificielle et une loi sur la gouvernance et la protection des données.

En Afrique du Sud, le ministère des Communications et des Technologies numériques a mis en place un groupe d’experts indépendant composé de sept membres afin d’examiner un projet de politique nationale en matière d’IA, à la suite du retrait de la version précédente en raison de références fictives et potentiellement générées par l’IA. Ce nouveau groupe a été chargé de réviser ou de supprimer les contenus problématiques et de corriger les citations ; la politique révisée devrait être approuvée par le Conseil des ministres d’ici novembre 2026 et faire l’objet d’une consultation publique prévue pour janvier 2027.

Par ailleurs, Singapour a mis à jour sa stratégie nationale en matière d’IA (NAIS), définissant 10 priorités actualisées couvrant la gouvernance, l’adoption par l’industrie, la recherche, les talents, le calcul, les données et la coopération internationale. Cette stratégie met l’accent sur la transformation sectorielle par le biais de missions nationales en matière d’IA, une intégration plus poussée de l’IA dans l’ensemble du secteur public, l’expansion de la recherche sur l’ensemble de la pile IA, et le développement de talents compétents en IA aux côtés de spécialistes. Elle met également l’accent sur une informatique économe en ressources, une gouvernance des données renforcée, et le positionnement de Singapour en tant que pôle de confiance pour la collaboration internationale en matière d’IA.

Cybersécurité

Les autorités du Royaume-Uni, de Singapour, d’Australie et de la Suisse ont averti que l’IA pourrait accélérer la découverte et l’exploitation des vulnérabilités logicielles, permettre de nouvelles formes d’attaques par apprentissage automatique adversaire et accroître l’ampleur et la sophistication des cyberopérations. Les régulateurs financiers du Royaume-Uni et d’Australie ont de même exhorté les organisations à renforcer la gouvernance, la gestion des vulnérabilités, la surveillance des tiers et les capacités de reprise, en prévision de l’émergence d’acteurs malveillants dotés d’une IA plus performante.

Les gouvernements réagissent en mettant en place de nouvelles institutions, des cadres réglementaires et des capacités de défense.

Les Émirats arabes unis ont lancé un laboratoire national de sécurité de l’IA axé sur la certification et la cyber-résilience, tandis que l’Australie a créé un institut de sécurité de l’IA chargé d’évaluer les systèmes d’IA émergents et de soutenir les régulateurs. Le Royaume-Uni a annoncé son intention de développer une capacité nationale de cyberdéfense alimentée par l’IA, basée sur des agents logiciels autonomes capables d’identifier et de répondre aux menaces à la vitesse d’une machine. L’Australie et le Royaume-Uni ont également renforcé la coopération entre leurs instituts respectifs de sécurité de l’IA, en mettant particulièrement l’accent sur les risques liés à l’IA de pointe et la cybersécurité.

Au niveau international, les efforts se concentrent de plus en plus sur les normes communes et la coordination. Le groupe de travail du G7 sur la cybersécurité a publié des lignes directrices pour une nomenclature logicielle de l’IA afin d’améliorer la transparence et la cybersécurité tout au long des chaînes d’approvisionnement de l’IA. Les ministres du Numérique et de la Technologie du G7 se sont engagés à faire progresser la coopération sur les cadres d’évaluation de l’IA, les lignes directrices en matière de marchés publics et la détection des contenus générés par l’IA.[1] Ils ont également appelé à des principes plus clairs sur l’ouverture de l’IA, à un soutien accru à la culture numérique et au développement des compétences en IA, ainsi qu’à des mesures ciblées pour aider les PME à adopter l’IA

L’intersection entre l’IA et la sécurité nationale devient également plus visible. La Corée du Sud a organisé un examen interministériel des cybermenaces liées à l’IA, réunissant des agences scientifiques, de défense et de renseignement afin de coordonner les réponses. Aux États-Unis, une cour d’appel fédérale a entendu les arguments d’Anthropic contestant sa désignation comme une menace pour la sécurité nationale au sein de la chaîne d’approvisionnement, soulignant les questions émergentes concernant la manière dont les gouvernements évaluent et réglementent les entreprises d’IA dans le cadre de la sécurité nationale.

Parallèlement, les capacités de l’IA s’étendent de plus en plus au-delà du domaine numérique. La Corée du Nord a annoncé des essais de missiles de croisière et de systèmes d’artillerie de précision pilotés par l’IA, marquant une nouvelle étape dans les efforts du pays pour moderniser ses forces armées traditionnelles. Les médias d’État ont indiqué que ces systèmes intègrent des technologies de ciblage assisté par l’IA et de navigation autonome conçues pour la guerre contemporaine.

Droits de l’Homme

Les autorités de protection de la vie privée examinent de plus en plus attentivement la manière dont les systèmes d’IA générative collectent, traitent et divulguent les données à caractère personnel, tandis que les mesures de contrôle et les recommandations se multiplient dans toutes les juridictions.

Les autorités fédérales et provinciales canadiennes chargées de la protection de la vie privée ont constaté que certains aspects de la collecte, de l’utilisation et de la divulgation de renseignements personnels par OpenAI via ChatGPT n’étaient pas conformes aux lois applicables en matière de protection de la vie personnelle dans le secteur privé. Les autorités ont estimé que la collecte initiale par l’entreprise de données à caractère personnel provenant de sites web accessibles au public et de sources tierces sous licence, à des fins d’entraînement des modèles, était trop large et donc inappropriée, compte tenu de l’ampleur, de la sensibilité et de l’inexactitude potentielle des données concernées, ainsi que des limites des mesures préventives en place à l’époque.

Le Royaume-Uni a mis en vigueur une réglementation exigeant que le commissaire à l’information élabore un code de bonnes pratiques concernant le traitement des données à caractère personnel en matière d’IA et de prise de décision automatisée. Ce code doit également inclure des recommandations sur les bonnes pratiques en matière de traitement des données à caractère personnel des enfants.

La CNIL, l’autorité française de protection des données, et la Commission sud-coréenne de protection des informations personnelles ont élaboré conjointement une affiche visant à sensibiliser le public aux risques liés à la vie privée dans le cadre de l’IA générative. Cette nouvelle affiche, intitulée « IA générative et vie privée », fournit des conseils pratiques sur la manière dont les utilisateurs peuvent protéger leurs données à caractère personnel avant, pendant et après l’utilisation de services d’IA générative.

Juridique

En Europe, les décideurs politiques ont continué à affiner l’architecture de mise en œuvre de la loi sur l’IA. La Commission européenne a publié un projet de lignes directrices sur les systèmes d’IA à haut risque, destiné à aider les fournisseurs et les exploitants à déterminer si les systèmes d’IA relèvent de la catégorie « à haut risque » prévue par la loi. Les institutions de l’UE sont également parvenues à un accord provisoire sur le dernier paquet omnibus, qui prolonge les délais de mise en œuvre, simplifie certaines obligations et renforce les pouvoirs du Bureau de l’IA. La Commission a également nommé de nouveaux organes d’experts, le groupe scientifique et le forum consultatif, afin de fournir une expertise indépendante au Bureau de l’IA de la Commission et aux autorités nationales chargées de superviser le respect du règlement.

L’Espagne a approuvé un projet de loi organique visant à intégrer la loi européenne sur l’IA dans son cadre juridique national. La proposition désigne l’agence nationale de surveillance de l’IA comme autorité centrale, tout en attribuant des responsabilités de surveillance sectorielles aux régulateurs financiers, de protection des données et judiciaires.

Les questions relatives à l’IA et à la propriété intellectuelle ont également retenu l’attention internationale. Au niveau de l’UE, la Commission européenne a lancé une consultation visant à déterminer s’il convient d’actualiser les règles en matière de droit d’auteur afin de tenir compte de l’évolution de l’économie numérique, y compris des défis posés par l’IA générative. Cette révision permettra d’évaluer l’efficacité de la directive de 2019 sur le droit d’auteur et d’examiner si les règles existantes restent adaptées à leur objectif, alors que les systèmes d’IA s’appuient de plus en plus sur des contenus protégés par le droit d’auteur. La Corée du Sud a publié un guide en anglais sur l’utilisation équitable de contenus protégés par le droit d’auteur pour l’entraînement de modèles d’IA générative et a fait part de son intention de jouer un rôle plus actif dans les discussions internationales sur l’application des exceptions au droit d’auteur au développement de l’IA. Parallèlement, l’Office de la propriété intellectuelle de Singapour et l’Office japonais des brevets ont lancé une initiative de coopération sur l’utilisation de l’IA dans l’examen des brevets, comprenant des échanges techniques et la formation des examinateurs visant à renforcer la qualité et la cohérence des processus d’examen des brevets

Parallèlement aux réformes réglementaires, les gouvernements élaborent également des cadres pour le déploiement de l’IA dans le secteur public. La Nouvelle-Zélande a publié des lignes directrices encourageant les régulateurs publics à adopter l’IA pour des tâches administratives à faible risque tout en maintenant un contrôle humain et une responsabilité. Ces lignes directrices mettent en avant les utilisations à faible risque, notamment le tri des dossiers, la hiérarchisation des priorités et la validation de données structurées.

Économie

Les institutions et les entreprises sont de plus en plus confrontées à la manière dont l’IA va remodeler les marchés du travail, les systèmes de compétences et les écosystèmes des plateformes, les divisions émergentes entre les politiques et l’industrie devenant de plus en plus visibles.

En matière de réglementation des plateformes, Meta a décidé de contester une décision de l’UE qui obligerait WhatsApp à permettre l’interopérabilité avec des chatbots IA concurrents. Cette affaire reflète la pression réglementaire croissante en Europe visant à ouvrir les plateformes de messagerie dominantes aux services IA tiers

Les déclarations publiques de hauts responsables d’Anthropic et d’OpenAI ont mis en évidence une divergence croissante au sein du secteur de l’IA quant à la vitesse et à l’ampleur de la transformation de la main-d’œuvre. S’exprimant lors d’une conférence au Vatican sur l’éthique de l’IA, Chris Olah, cofondateur d’Anthropic, a averti que le remplacement à grande échelle des cols blancs reste un scénario plausible à mesure que les systèmes de pointe progressent. Sam Altman, PDG d’OpenAI, a adopté une position plus prudente, affirmant que l’IA a jusqu’à présent eu un impact plus limité que prévu sur l’emploi des cols blancs débutants, suggérant que les prévisions antérieures d’une disruption rapide ont peut-être surestimé le rythme du changement.

Dans le même temps, la Corée du Sud a mis en place un comité tripartite sur l’IA et le travail, réunissant le gouvernement, les employeurs et les travailleurs afin d’évaluer les implications de l’adoption de l’IA sur l’emploi et de coordonner les réponses politiques. Cette initiative s’inscrit dans un effort plus large visant à anticiper les bouleversements du marché du travail et à structurer le dialogue entre les principaux acteurs économiques. L’Organisation internationale du travail a également averti que l’apprentissage tout au long de la vie serait essentiel pour la future économie de l’IA, soulignant la nécessité d’un développement continu des compétences à mesure que l’automatisation et l’IA redessinent les structures professionnelles.

Socioculturel

Les organisations internationales et les régulateurs redoublent d’efforts pour intégrer des applications éthiques et socialement bénéfiques de l’IA dans les cadres culturels, éducatifs et de gouvernance.

Le Bureau régional de l’UNESCO pour l’Asie de l’Est a lancé un appel mondial à des exemples de bonnes pratiques sur la manière dont l’IA et le design sont utilisés pour soutenir la culture, la créativité, l’éducation, la durabilité et l’inclusion sociale. Cet appel invite les organisations, institutions, praticiens, éducateurs et innovateurs qui utilisent l’IA en association avec des approches de design pour créer des résultats positifs dans les secteurs culturels et créatifs à soumettre leurs contributions.

L’UNESCO a annoncé le lancement de sa méthodologie d’évaluation de l’état de préparation à l’IA au Kazakhstan afin d’évaluer l’état de préparation du pays en matière de gouvernance et de développement de l’IA. Ce cadre vise à aider les pays à aligner leurs approches de gouvernance de l’IA sur la Recommandation de l’UNESCO sur l’éthique de l’IA.

Un programme pilote national d’évaluation éthique et de services liés à l’IA a été lancé en Chine, alors que les autorités s’efforcent de renforcer la surveillance des risques croissants liés aux systèmes d’IA avancés. Cette initiative sera dans un premier temps mise en œuvre dans certaines zones d’innovation provinciales et visera principalement à développer des mécanismes d’évaluation éthique, à mettre en place des comités spécialisés et à créer des centres de services dédiés à l’éthique. Les autorités de régulation prévoient également d’intégrer les processus d’évaluation éthique dans des normes techniques et de renforcer les mécanismes de signalement des problèmes éthiques liés à l’IA.

Gouvernance des contenus

Les gouvernements et les institutions adoptent de plus en plus de règles et de lignes directrices ciblées pour encadrer l’utilisation de l’IA générative dans les médias, la recherche, l’éducation et les services publics.

Le Vietnam a mis en place de nouvelles obligations de divulgation pour les contenus générés et modifiés par l’IA, imposant des mentions claires pour les fichiers audio, les images et les vidéos qui imitent des personnes réelles ou recréent de manière convaincante des événements du monde réel, dans le cadre d’efforts plus larges visant à améliorer la transparence et à faire face aux risques associés à des médias synthétiques de plus en plus réalistes.

La Commission européenne a mis à jour ses lignes directrices « ERA Living Guidelines » sur l’utilisation responsable de l’IA générative dans la recherche, en introduisant de nouvelles recommandations concernant la confidentialité, la propriété intellectuelle et les risques liés à la protection des données dans le cadre de l’utilisation d’outils d’IA externes, tout en soulignant les menaces émergentes telles que les messages cachés conçus pour en manipuler ses systèmes . La Commission européenne a également révisé ses recommandations sur l’utilisation éthique de l’IA et des données dans l’éducation afin d’aider les enseignants et les responsables scolaires à utiliser cette technologie de manière sûre et responsable, conformément aux valeurs de l’UE.

Par ailleurs, les organisations médiatiques suisses ont adopté un code de conduite national pour l’utilisation de l’IA dans le journalisme, renforçant la responsabilité, la transparence et la protection des droits d’auteur.

Health New Zealand a publié des lignes directrices appelant à la prudence dans l’utilisation de l’IA générative et des grands modèles linguistiques dans les établissements de santé, en limitant leur utilisation pour les décisions cliniques et le traitement de données sensibles.

Renforcement des capacités

Le Conseil de l’Union européenne a approuvé des conclusions appelant à une approche éthique, sûre et centrée sur l’humain en matière d’IA dans l’éducation, soulignant que les enseignants doivent rester au cœur du processus d’apprentissage alors que les outils d’IA sont de plus en plus utilisés dans les écoles et les universités. Ces conclusions invitent les gouvernements nationaux à renforcer les compétences des enseignants en matière d’IA et de numérique par le recours à la formation, tout en encourageant le développement et l’utilisation d’outils d’IA spécifiques à l’éducation qui apportent une valeur pédagogique avérée et respectent les exigences en matière de protection des données, de responsabilité et de sensibilisation aux risques.

Parallèlement, la Chine a lancé une plateforme mondiale de services éducatifs en IA afin d’élargir l’accès transfrontalier aux ressources d’apprentissage numériques et de soutenir l’intégration de l’IA dans l’éducation. La plateforme est désormais accessible dans environ 220 pays et régions.

Par ailleurs, plusieurs pays se concentrent sur la culture numérique en matière d’IA au niveau national et sur son adoption par le grand public. Malte a lancé « AI for All », un programme national de sensibilisation à l’IA destiné à aider les citoyens et les résidents à comprendre les capacités, les limites et l’utilisation responsable de l’IA dans la vie quotidienne et au travail, parallèlement à un partenariat offrant l’accès à des outils d’IA avancés. La Grèce a publié « Artificial Intelligence for All », un guide public sur l’IA et les grands modèles linguistiques visant à améliorer la compréhension de base du fonctionnement de ces systèmes et de leur utilisation concrète. Le Centre national australien de l’IA a lancé AI.gov.au, une plateforme centrale destinée à soutenir une adoption sûre et responsable de l’IA au sein des entreprises et des organisations à but non lucratif.

Les gouvernements considèrent de plus en plus l’IA comme un outil structurel au service de la modernisation de l’État. Par exemple, le ministère argentin du Capital humain a lancé l’initiative « Digital Twin », un système basé sur l’IA destiné à simuler les impacts potentiels des politiques sociales avant leur mise en œuvre. Ce système est conçu pour modéliser des scénarios liés à des domaines tels que la pauvreté, les subventions et le développement du capital humain à l’aide d’ensembles de données à grande échelle. Selon les responsables, ce projet s’inscrit dans le cadre d’efforts plus larges visant à utiliser l’analyse de données et des outils prédictifs dans la planification des politiques publiques. La Nouvelle-Zélande a présenté ses projets pour une réforme majeure du secteur public qui place l’IA au cœur de la modernisation de l’administration et de la prestation de services. Le Kazakhstan a examiné des propositions visant à étendre le déploiement de l’IA à tous les secteurs dans le cadre de son programme de transformation numérique et prévoit d’accroître la capacité informatique nationale, d’améliorer les systèmes d’allocation des ressources afin d’éviter les goulots d’étranglement au niveau des infrastructures, et de renforcer l’infrastructure de données grâce à des initiatives telles que la « Data Center Valley » à Pavlodar, parallèlement à une mise à jour de la réglementation en matière de gouvernance du cloud.Les pays investissent également dans le renforcement de leurs capacités structurelles. Le Canada renforce ses capacités nationales en photonique et en semi-conducteurs grâce à la commercialisation du Centre canadien de fabrication en photonique du Conseil national de recherches. En Europe, l’Université de Belgrade, en Serbie, a annoncé le lancement de SAIFA, une initiative d’infrastructure d’IA soutenue par l’entreprise commune EuroHPC, visant à élargir l’accès au calcul haute performance pour la recherche, l’industrie et l’administration publique.

Magnifica humanitas : la dignité humaine à l’ère de l’IA

Le pape Léon XIV a publié Magnifica Humanitas, une encyclique novatrice qui expose la position de l’Église catholique sur l’intelligence artificielle.

Pour en saisir toute la portée, il convient de rappeler qu’une encyclique est une lettre pastorale rédigée par le pape à l’intention de l’ensemble de l’Église catholique romaine sur des questions de doctrine, de morale ou de discipline. Elle s’adresse également au-delà de l’Église, influençant souvent les débats éthiques et politiques mondiaux sur des questions majeures telles que les inégalités, la guerre, l’environnement et la technologie.

Lorsque des détracteurs s’opposèrent à la publication par le pape Léon XIII de Rerum Novarum sur la société, l’économie et la politique, le pape fit remarquer que la proclamation de l’Évangile ne pouvait ignorer la vie concrète des gens. C’est cette même logique qui sous-tend Magnifica Humanitas.

Adult, Male, Man, Person, Clothing, Coat, Face, Head, Animal, Bird

Ancrée dans des millénaires d’histoire, Magnifica Humanitas offre un antidote important au discours sur l’IA présenté comme une fatalité : la course ne peut être arrêtée, le marché ne peut être ralenti, la concurrence géopolitique ne peut être évitée, et les humains doivent simplement suivre. L’encyclique rejette cette fatalité. Elle insiste sur le fait que l’avenir technologique doit être défini par nos choix.

L’encyclique définit la gouvernance de l’IA à travers deux métaphores contrastées : la tour de Babel et les murs de la Nouvelle Jérusalem. Babel incarne l’ambition technologique dépourvue d’humilité, qui s’élève vers le ciel par l’ambition, le contrôle et l’exaltation de soi, pour finalement n’engendrer que fragmentation, confusion et effondrement. La métaphore des murs de Jérusalem offre une vision différente. Elle se construit pièce par pièce, grâce à la coopération, à la solidarité, au dialogue et à la responsabilité. De plus, elle requiert la diplomatie plutôt que la domination, la patience plutôt que l’accélération, et le bien commun plutôt que la course à la suprématie.

En fin de compte, « Magnifica Humanitas » se résume à un choix entre deux métaphores : Babel et Jérusalem. L’encyclique nous met en garde contre le risque de sombrer, comme des somnambules, dans Babel, et nous offre des repères pour atteindre Jérusalem en puisant dans les forces cognitives, émotionnelles et spirituelles qui font de nous des êtres humains.

Comment réparer le « système d’exploitation » de l’humanité. L’encyclique présente le « système d’exploitation » de l’humanité comme ancré dans les traditions de l’Âge axial qui ont défini l’identité morale, rationnelle et spirituelle de l’homme, puis affinées par la pensée des Lumières. Elle met en garde contre le fait que le transhumanisme et le posthumanisme, portés par l’IA et la bio-ingénierie, risquent de redéfinir les humains comme des entités à optimiser plutôt que comme des êtres dotés d’une dignité intrinsèque. Magnifica Humanitas se concentre sur ce dilemme central, en écartant tous les artifices des discours technologiques, commerciaux et autres qui dominent notre espace de communication. Elle appelle également à revisiter le cœur du « système d’exploitation » de l’humanité, centré sur la vie et la dignité des êtres humains, en soutenant que les humains ne devraient pas être « optimisés » de la même manière que les technologies.

L’encyclique met en garde contre le danger de considérer les systèmes d’IA comme neutres et objectifs, soulignant que la technologie prend les caractéristiques de ceux qui la conçoivent, la financent, la réglementent et l’utilisent.

Clarté sur la gouvernance et la responsabilité en matière d’IA. En ce qui concerne la gouvernance de l’IA, l’encyclique apporte des éclaircissements. Elle remet en question le paradigme technocratique, dans lequel la gouvernance est traitée comme un exercice d’expertise axé principalement sur les normes, la gestion des risques et l’efficacité. Au contraire, l’encyclique soutient que la gouvernance de l’IA doit partir de la dignité humaine et du bien commun. Les décisions concernant l’IA devraient être prises par les personnes et les communautés directement concernées. Les communautés ne devraient pas se contenter d’être consultées après que les règles et les plateformes ont été définies ailleurs. Elles doivent être en mesure d’influencer, de remettre en question et de corriger les systèmes qui ont un impact sur leur vie. Cela fait fortement écho à la tradition multipartite de la gouvernance de l’internet et remet en cause à la fois le monopole des entreprises et le contrôle centré sur l’État.

Un regard neuf sur les droits de l’Homme. L’encyclique met en évidence un paradoxe de notre époque : les droits de l’Homme sont largement proclamés de manière formelle, mais rapidement érodés, entre autres, par le progrès technologique. À travers le profilage, la manipulation et l’optimisation technologique globale, les personnes sont de plus en plus traitées comme des points de données et des éléments de modèles algorithmiques. Par conséquent, l’une des contributions les plus essentielles de l’encyclique est son ancrage renouvelé des droits de l’Homme pour protéger notre humanité fondamentale à l’ère de l’IA.

Désarmer l’IA. Un concept marquant de l’encyclique est l’appel à « désarmer l’IA ». Cela ne se limite pas à la réglementation des armes autonomes ou des systèmes militaires, bien que ces questions soient urgentes. Il s’agit d’un appel plus large au désarmement cognitif, économique et politique. L’IA devient un instrument de rivalité géopolitique, de domination commerciale, de surveillance, de propagande et de contrôle social. L’encyclique remet en cause l’idée selon laquelle la course à l’IA est inévitable. Elle appelle à ce que l’IA soit développée comme un outil au service de l’humanité plutôt que comme une arme de pouvoir et de guerre.

L’esclavage et le colonialisme à l’ère de l’IA. Le colonialisme numérique est une autre préoccupation de l’encyclique. Les puissances coloniales n’ont plus besoin de contrôler des territoires. Elles peuvent s’approprier des données, influencer les marchés, contrôler les infrastructures et extraire de la valeur de vies humaines transformées en informations exploitables. Un ordre numérique véritablement postcolonial rendrait l’autonomie aux individus et aux communautés. Il donnerait aux personnes non seulement l’accès à leurs données, mais aussi un contrôle significatif sur la manière dont ces données sont utilisées, par qui et au profit de qui.

Comment faire face aux risques liés à l’IA. Magnifica Humanitas offre une analyse éclairée et percutante des risques liés à l’IA. Concernant le risque existentiel, elle s’écarte des discours centrés sur la prise de pouvoir par l’IA générale (AGI) ou la super-intelligence, arguant au contraire que les risques existentiels peuvent émerger plus progressivement, et pas nécessairement de la technologie elle-même, mais de ceux qui possèdent et exploitent les systèmes d’IA. Concernant le risque d’exclusion, il met en garde contre la concentration du pouvoir entre les mains de quelques-uns et l’exclusion de nombreux autres, les monopoles prenant des formes épistémiques, économiques et politiques. Il souligne comment l’opacité et l’absence de contrôle public peuvent engendrer des dépendances, des exclusions, des manipulations et des inégalités, notamment par le contrôle de la visibilité, de l’amplification et de l’influence comportementale. Il examine également les risques existants en matière d’emploi, d’économie, de désinformation et d’éducation.

Langue et culture à l’ère de l’IA. Le Pape met en garde contre le fait que la violence commence souvent par le langage : les guerres sont précédées par des guerres de mots. À cet égard, l’IA joue un rôle crucial dans l’influence de la culture en tant que génératrice de textes, de vidéos et d’autres artefacts de communication. L’encyclique appelle donc à accorder une attention renouvelée à la vérité, au dialogue et à l’écologie morale du langage. L’IA présente les risques principaux de générer des récits synthétiques, une persuasion superficielle et une distorsion automatisée. Le défi ne réside pas seulement dans la désinformation. Il s’agit de l’érosion progressive de la réalité partagée.

L’éducation à l’ère de l’instantanéité. Magnifica Humanitas accorde une grande importance à l’éducation à l’ère de l’IA, conformément à la longue tradition doctrinale de l’Église catholique. L’encyclique met en évidence une tension fondamentale entre l’accès rapide à l’information à l’ère numérique et la nature intrinsèquement lente de l’apprentissage. L’information peut être transmise instantanément. La connaissance, et plus encore la sagesse, exigent beaucoup de temps. Or, les écoles et les universités ne sont pas préparées à gérer cette tension entre la technologie de l’IA et l’apprentissage humain.

Impact de l’IA sur la nature et l’environnement. L’encyclique nous rappelle également que rien dans le monde numérique n’est immatériel ou magique. L’IA dépend d’infrastructures physiques : centres de données, réseaux énergétiques, minéraux, eau, appareils, câbles et chaînes d’approvisionnement. Les systèmes d’IA actuels nécessitent d’énormes quantités d’énergie et d’eau et exercent une pression croissante sur les ressources naturelles. Le coût environnemental de l’IA n’est donc pas un élément extérieur au débat. Une technologie qui prétend faire progresser l’humanité ne peut être évaluée uniquement en fonction de sa rapidité, de sa précision ou de sa rentabilité. Elle doit également être jugée à l’aune de son impact sur la planète, sur les communautés vulnérables et sur les générations futures.

Pourquoi Magnifica Humanitas est importante. L’importance historique de Magnifica Humanitas réside dans sa tentative de faire passer le débat sur l’IA de la gestion technique à une réflexion civilisationnelle. Elle remet en cause la concentration du pouvoir entre les mains de quelques entreprises. Elle remet en question les discours transhumanistes et posthumanistes qui traitent la personne humaine comme obsolète ou perfectible. Elle insiste sur le fait que la gouvernance de l’IA doit être fondée sur la dignité, la subsidiarité, la solidarité, la justice et le bien commun. Elle offre de nouvelles perspectives sur les risques d’esclavage numérique, de colonialisme, de militarisation, de manipulation et de régression anthropologique.

Plus important encore, il rétablit l’autonomie morale. Métaphoriquement parlant, nous ne sommes pas condamnés à construire une tour de Babel. Nous pouvons choisir de bâtir les murs de la nouvelle Jérusalem.Ce texte est une adaptation de l’analyse du Dr Jovan Kurbalija sur Magnifica Humanitas, qui sera développée dans les prochains jours. Pour recevoir une analyse approfondie, vous pouvez vous inscrire ici.

Au-delà de l’IA : les développements qui ont fait sensation en mai

Technologies

La Norvège a rejoint l’initiative Pax Silica, qui se concentre sur la sécurité des chaînes d’approvisionnement en IA, en semi-conducteurs et en matières premières critiques. Les responsables norvégiens ont déclaré que cette participation pourrait améliorer l’accès au marché pour les entreprises nationales opérant dans les secteurs technologiques de pointe et renforcer la coopération en matière de sécurité économique avec des partenaires stratégiques.

La Commission européenne a adopté une proposition concernant la future attribution de la bande de fréquences de 2 GHz pour les services mobiles par satellite de l’UE au-delà de 2027. Ce spectre est de plus en plus considéré comme un atout stratégique tant pour la connectivité commerciale que pour les communications liées à la défense, en particulier pour les services satellitaires de communication directe entre appareils (D2D) et les infrastructures de communication critiques. Selon cette proposition, un tiers du spectre serait réservé à un usage gouvernemental, sécuritaire et militaire, dans le cadre du programme de connectivité sécurisée IRIS² de l’UE. Les deux tiers restants seraient destinés aux services commerciaux, mais la moitié de cette capacité serait réservée spécifiquement aux opérateurs de l’UE entrant sur le marché, tandis que le reste resterait ouvert aux entreprises tant de l’UE que hors UE.

Cybersécurité

L’UE a adopté des modèles communs pour la notification des incidents de cybersécurité à l’échelle de l’Union afin de réduire les charges administratives et de simplifier la mise en conformité pour les entreprises tenues de signaler les incidents de cybersécurité en vertu de la directive NIS2. La Commission prévoit désormais d’adopter ces modèles par le biais d’un acte d’exécution, ce qui les rendrait obligatoires pour tous les États membres.

Le Conseil de l’Union européenne a prolongé jusqu’au 18 mai 2027 les mesures restrictives à l’encontre des personnes et entités impliquées dans des cyberattaques menaçant l’UE et ses États membres. Ce cadre permet à l’UE d’imposer des sanctions ciblées à l’encontre de personnes ou d’entités impliquées dans des cyberattaques significatives constituant une menace extérieure pour l’Union ou ses États membres. Des mesures peuvent également être imposées en réponse à des cyberattaques contre des pays tiers ou des organisations internationales, lorsqu’elles soutiennent les objectifs de la politique étrangère et de sécurité commune.

Le gouvernement fédéral allemand a approuvé un projet de loi visant à renforcer les capacités de cyberdéfense de trois agences fédérales : l’Office fédéral pour la sécurité de l’information (BSI), l’Office fédéral de police criminelle (BKA) et la police fédérale (Bundespolizei). Les autorités seraient en mesure de bloquer ou de perturber les logiciels et les infrastructures de serveurs utilisés dans le cadre de cyberattaques, y compris les systèmes situés en dehors de l’Allemagne. Le BSI se verrait également conférer des pouvoirs élargis pour collecter, stocker et analyser des données afin de détecter des activités indiquant la préparation d’une attaque.

YouTube, Snap, TikTok et Meta ont accepté de conclure un accord à l’amiable dans le cadre d’un procès historique intenté par le district scolaire du comté de Breathitt, dans le Kentucky, qui les accusait d’avoir conçu des systèmes algorithmiques et des fonctionnalités — telles que le défilement infini et les boucles de recommandation axées sur l’engagement — qui favorisent une utilisation compulsive chez les jeunes utilisateurs et imposent des charges financières et pédagogiques supplémentaires aux établissements scolaires. Les termes de l’accord n’ont pas été divulgués. Le procès devait s’ouvrir le 12 juin, mais il n’aura finalement pas lieu.

Parallèlement, l’Ofcom britannique a indiqué que la quasi-totalité des jeunes âgés de 8 à 17 ans au Royaume-Uni sont connectés à Internet (99 %), ce qui le conforte dans sa volonté de renforcer la responsabilité des plateformes. L’Ofcom souhaite que la sécurité des enfants soit intégrée dès le départ dans la conception des produits, et que les régulateurs soient informés à l’avance des nouvelles fonctionnalités afin d’évaluer les risques potentiels. Toutefois, l’autorité de régulation reste préoccupée par le fait que certaines plateformes ne font pas assez d’efforts pour rendre les flux plus sûrs et doute que les politiques fixant l’âge minimum à 13 ans empêchent efficacement l’accès des mineurs.

Le Commissariat à la protection de la vie privée du Canada a publié des lignes directrices sur la manière dont les organisations doivent évaluer et mettre en œuvre des outils de vérification de l’âge pour les sites Web et les services en ligne. Le CPVP précise que la vérification de l’âge ne doit être utilisée que lorsqu’il existe une obligation légale claire ou un risque avéré de préjudice pour les enfants. Il souligne que les organisations doivent déterminer si des mesures alternatives moins intrusives pourraient permettre de contrer ces risques avant d’adopter de tels systèmes.

La Commission des communications et du multimédia de Malaisie (MCMC) a publié le Code de protection de l’enfance et le Code d’atténuation des risques en vertu de la loi de 2025 sur la sécurité en ligne. Les nouvelles règles exigent des mesures de protection plus strictes pour les jeunes utilisateurs sur les plateformes en ligne, notamment des contrôles d’âge plus rigoureux et des obligations renforcées en matière de gouvernance des contenus pour les fournisseurs de services. La Malaisie a également commencé à interdire aux enfants de moins de 16 ans de créer des comptes sur les réseaux sociaux. Les grandes plateformes telles que Facebook, Instagram, TikTok et YouTube sont tenues de vérifier l’âge des utilisateurs par rapport aux registres gouvernementaux, sous peine de sanctions pouvant atteindre 10 millions de ringgits (environ 2,5 millions de dollars américains) en cas de non-respect. Les utilisateurs existants feront l’objet de contrôles de vérification d’âge pendant une période de transition de six mois.

Par ailleurs, la Suède envisage également d’interdire l’accès aux réseaux sociaux aux jeunes utilisateurs. Dans un rapport intermédiaire soumis au ministre des Affaires sociales, les enquêteurs gouvernementaux nommés l’année dernière ont recommandé d’interdire aux moins de 15 ans l’accès aux plateformes permettant aux utilisateurs de se découvrir, d’entrer en contact et de communiquer entre eux, ou sur lesquelles du contenu peut être partagé ou découvert.

Les ministres du G7 chargés du numérique et de la technologie se sont mis d’accord sur un ensemble de principes communs définissant un espace numérique plus sûr et plus sécurisé pour les mineurs. Les ministres mettent l’accent sur une vérification d’âge efficace, des fonctionnalités de sécurité intégrées dès la conception des plateformes, notamment des protections par défaut, des contrôles parentaux et des systèmes de recommandation limitant l’exposition à des contenus préjudiciables ou addictifs. Ils appellent également à une prévention et une répression rigoureuses contre les préjudices graves tels que les contenus pédopornographiques et les images intimes non consenties, y compris celles générées par l’IA, ainsi qu’à un meilleur soutien aux victimes. Parallèlement, ils mettent l’accent sur la mise à disposition des parents d’outils de contrôle faciles à utiliser, le renforcement des compétences numériques et en matière d’IA chez les mineurs, l’amélioration de la transparence et du contrôle des utilisateurs sur leurs données, et la garantie que les plateformes évaluent, atténuent et signalent systématiquement les risques. Ces principes soulignent également la nécessité d’une action coordonnée entre les gouvernements, l’industrie, les chercheurs, les éducateurs et la société civile.

Droits de l’Homme

Depuis le 8 mai, la messagerie chiffrée de bout en bout sur Instagram n’est officiellement plus disponible. Meta a désactivé cette fonctionnalité à l’échelle mondiale, renonçant ainsi à son projet d’étendre cette technologie de protection de la vie privée à l’ensemble de la plateforme, après avoir passé des années à promouvoir la communication chiffrée comme l’avenir de la messagerie.

Apple et Meta s’opposent au projet de loi C-22 proposé par le Canada, qui, selon eux, pourrait contraindre les entreprises à affaiblir le chiffrement ou à intégrer des mécanismes d’accès gouvernemental dans leurs produits. Les autorités canadiennes affirment que ce projet de loi aiderait les forces de l’ordre à réagir plus rapidement aux menaces pour la sécurité.

Après 88 jours de coupure à l’échelle nationale, la connectivité Internet a été partiellement rétablie en Iran, mettant fin à l’une des plus longues coupures totales de l’histoire moderne. Cette coupure avait été imposée suite à l’escalade du conflit et justifiée par les autorités pour des raisons de sécurité. NetBlocks a indiqué que la connectivité avait largement été rétablie, mais que les indicateurs montrent que les utilisateurs sont toujours confrontés à un filtrage intensif, notamment de nouvelles restrictions sur la messagerie et les boutiques d’applications.

Le président russe Vladimir Poutine a donné pour instruction au gouvernement et au FSB de garantir un accès ininterrompu aux services médicaux, d’information et de paiement essentiels pendant les périodes de restriction de l’Internet mobile en Russie. Le Premier ministre Mikhaïl Michoustine et le directeur du FSB Alexandre Bortnikov doivent rendre compte de leurs progrès d’ici le 1er juillet.

Juridique

La Commission européenne a présenté un paquet de mesures sur la souveraineté technologique européenne visant à renforcer les capacités de l’Europe dans les domaines des semi-conducteurs, de l’IA, des infrastructures cloud et des technologies open source. Ce paquet comprend deux propositions législatives, la loi sur les puces 2.0 et la loi sur le développement du cloud et de l’IA, ainsi qu’une stratégie open source et une feuille de route stratégique pour la numérisation et l’IA dans le secteur de l’énergie. La proposition de « Chips Act 2.0 » vise à développer les capacités de l’Europe en matière de semi-conducteurs et à soutenir la prochaine génération d’infrastructures informatiques nécessaires à l’IA et à d’autres technologies de pointe. La proposition de « Cloud and AI Development Act » se concentre sur le renforcement des infrastructures européennes de cloud et d’IA, y compris la capacité des centres de données, alors que la demande en ressources informatiques augmente avec la généralisation de l’IA. Parallèlement à ces propositions législatives, la Commission a présenté une stratégie en matière de logiciels libres visant à soutenir l’écosystème logiciel européen et à réduire la dépendance vis-à-vis des technologies fermées ou contrôlées de l’extérieur. La feuille de route stratégique pour la numérisation et l’IA dans le secteur de l’énergie vise à soutenir l’utilisation d’outils numériques et de l’IA dans ce secteur tout en contribuant à un avenir numérique plus durable.

La Chine a mis à jour ses règles de protection des secrets d’affaires afin d’y inclure explicitement les actifs numériques tels que les données, les algorithmes, le code logiciel et les programmes informatiques. Ce cadre renforce les exigences en matière de mesures de protection des entreprises, notamment les accords de confidentialité, la formation des employés, les contrôles d’accès, le chiffrement et les pratiques de gestion des données. Il introduit également des dispositions détaillées pour les environnements de travail numériques et à distance, encourageant des outils tels que l’accès basé sur des autorisations, le masquage des données et la journalisation des activités, tout en interdisant explicitement les méthodes illicites telles que le piratage, la fraude et l’accès non autorisé aux systèmes pour obtenir des secrets d’affaires. Ces règles étendent en outre la protection aux infractions commises à l’étranger qui affectent le marché intérieur chinois ou les droits des entreprises chinoises.

Meta Platforms fait face à une pression juridique et réglementaire croissante tant aux États-Unis qu’en Europe, suite à des allégations selon lesquelles ses plateformes de réseaux sociaux contribueraient à la dépendance des jeunes et à des problèmes de santé mentale. Au Nouveau-Mexique, l’État réclame 3,7 milliards de dollars et demande au tribunal de déclarer Meta comme une nuisance publique. Le procès allègue que Facebook, Instagram et WhatsApp ont été conçus de manière à encourager un comportement addictif chez les mineurs. Il affirme également que ces plateformes n’ont pas protégé de manière adéquate les jeunes utilisateurs contre les contenus préjudiciables et l’exploitation. L’État demande des changements majeurs aux plateformes, notamment la vérification de l’âge et des restrictions sur des fonctionnalités telles que la lecture automatique et le défilement infini pour les mineurs. Meta affirme que l’affaire concerne des utilisateurs individuels, plutôt que le préjudice causé au public dans son ensemble.

Meta tente également de faire annuler le verdict d’un jury californien qui a jugé l’entreprise négligente dans la conception de ses plateformes et a accordé des dommages-intérêts à une jeune plaignante qui affirmait que l’utilisation des réseaux sociaux avait contribué à sa dépression. Meta fait valoir que ces plaintes sont irrecevables en vertu de l’article 230 du Communications Decency Act et que les préjudices allégués étaient liés au contenu en ligne plutôt qu’aux fonctionnalités de conception des plateformes.

À Milan, des familles et le MOIGE ont intenté une action contre Meta et TikTok pour manquement présumé à l’obligation de protéger les mineurs sur les plateformes de réseaux sociaux.

Le procureur général du Texas a intenté une action en justice contre WhatsApp et sa société mère, Meta Platforms Inc, alléguant que les utilisateurs auraient reçu des informations trompeuses sur la portée et l’efficacité des protections offertes par le chiffrement de bout en bout de WhatsApp. La plainte allègue que Meta et WhatsApp ont conservé un accès plus large à certaines communications et métadonnées des utilisateurs que ce que ces derniers auraient pu comprendre.

Le Texas a poursuivi Netflix pour avoir prétendument collecté des données d’utilisateurs sans leur consentement, notamment leurs habitudes de visionnage, et pour avoir transformé l’activité des utilisateurs en données monétisées. Le procureur général fait également valoir que des fonctionnalités telles que la lecture automatique ont été conçues pour accroître l’engagement, en particulier chez les mineurs, et que ces pratiques sont en contradiction avec les déclarations publiques de Netflix selon lesquelles la plateforme serait sans publicité et adaptée à toute la famille.

La Cour de justice de l’Union européenne a confirmé une décision italienne obligeant Meta à indemniser les éditeurs pour les extraits d’actualités, intensifiant ainsi les litiges en matière de droits d’auteur concernant la réutilisation de contenus et les données d’entraînement de l’IA.

La Cour de cassation française a rejeté le recours d’Amazon contre un prix minimum de livraison de livres destiné à protéger les librairies indépendantes contre la domination des plateformes mondiales.

La Comisiun na Mean a ouvert deux enquêtes sur Meta en vertu de la loi sur les services numériques (DSA), afin de déterminer si Facebook et Instagram permettent aux utilisateurs de choisir effectivement des flux de recommandations qui ne sont pas basés sur le profilage de leurs données personnelles. L’autorité de régulation a également averti que les très grandes plateformes en ligne (VLOP) doivent veiller à ce que les utilisateurs puissent exercer leurs droits en vertu de la DSA sans manipulation ni obstacles inutiles.

Un jury fédéral de Californie s’est prononcé en faveur d’OpenAI, de Sam Altman et de Greg Brockman dans le procès intenté par Elon Musk, qui alléguait que la société avait abandonné sa mission initiale à but non lucratif après être passée à une structure à but lucratif. Le tribunal a accepté l’argument d’OpenAI selon lequel Musk était depuis longtemps au courant des discussions concernant la restructuration, concluant que les plaintes avaient été déposées hors des délais légaux applicables et rejetant l’affaire à la suite d’un verdict consultatif du jury.

La Commission européenne a infligé une amende de 200 millions d’euros à Temu après avoir constaté que l’évaluation des risques de Temu pour 2024 ne répondait pas aux normes énoncées dans la loi sur les services numériques (DSA). La Commission a en outre estimé que Temu n’avait pas évalué de manière adéquate comment la conception de sa plateforme, en particulier les systèmes de recommandation et la promotion par des influenceurs, pouvait contribuer à l’amplification et à la diffusion de produits illégaux.

Socioculturel

L’Ofcom a annoncé une recommandation visant à renforcer les protections contre l’utilisation abusive d’images intimes illégales en ligne. L’autorité de régulation britannique a déclaré qu’elle mettait à jour ses codes relatifs aux contenus illégaux afin de recommander à certaines plateformes en ligne d’utiliser des technologies de détection automatisée pour identifier ces images, y compris les deepfakes explicites générés par l’IA et le partage d’images sans consentement. La recommandation devrait entrer en vigueur à l’automne 2026, sous réserve de l’approbation du Parlement.

De même, la Commission fédérale du commerce des États-Unis (FTC) a publié des lignes directrices à l’intention des plateformes en ligne concernant le respect de l’article 3 de la loi « Take It Down Act », entrée en vigueur le 19 mai 2026, qui impose aux plateformes concernées de retirer les photos ou vidéos intimes non consenties dans les 48 heures suivant la réception d’une demande valide. Les plateformes concernées doivent fournir des informations claires, bien visibles et rédigées dans un langage simple sur la manière dont les personnes peuvent soumettre des demandes de retrait concernant des photos ou vidéos intimes partagées sans consentement. Les plateformes doivent rendre la procédure facile à utiliser, y compris pour les personnes qui ne disposent pas d’un compte sur le service. La FTC encourage également les plateformes à contribuer à empêcher la diffusion des images supprimées, notamment grâce à la technologie de hachage et, le cas échéant, en partageant les hachages avec des services tels que le service « Take It Down » du Centre national pour les enfants disparus et exploités (National Center for Missing and Exploited Children) ou StopNCII.org.

Le ministère des Technologies de l’information et des communications de Papouasie-Nouvelle-Guinée a commencé à rédiger des instructions et un projet de loi sur l’identité numérique et les identifiants vérifiables, à la suite de l’adoption de la politique nationale en matière d’identité numérique. Le projet de loi soutiendra le déploiement national de SevisPass, SevisWallet, SevisDEx et d’autres identifiants vérifiables approuvés. SevisWallet permettra aux citoyens d’enregistrer, de conserver et de présenter des identifiants numériques fiables, tandis que SevisDEx permettra un échange de données sécurisé et fondé sur le consentement.

Des citoyens néerlandais et des défenseurs de la vie privée ont contesté le projet d’acquisition de Solvinity, la société hébergeant le système d’identité DigiD des Pays-Bas, par la société américaine Kyndryl. Les détracteurs ont averti que le fait de laisser une entreprise américaine approcher une infrastructure aussi sensible pourrait exposer les données des citoyens néerlandais à une juridiction étrangère et à des risques de surveillance en vertu des lois américaines. Au début du mois de mai, le tribunal de district de La Haye a rejeté une tentative de trois citoyens néerlandais visant à empêcher le gouvernement de renouveler son contrat avec Solvinity. Cependant, le Bureau d’évaluation des investissements (BTI) a conseillé à Willemijn Aerdts, ministre de l’Économie numérique et de la Souveraineté des Pays-Bas, d’imposer une interdiction totale de cette acquisition, car elle pourrait présenter un risque pour l’intérêt public. La ministre a suivi ce conseil, et le gouvernement a bloqué la vente.

L’UNESCO et le Pakistan ont lancé « Digital Citizens for Peace », une initiative d’éducation aux médias visant à lutter contre les discours de haine et la désinformation en formant de jeunes journalistes et créateurs de contenu. Grâce à des camps immersifs d’éducation aux médias et à l’information, des programmes de mentorat et des kits pédagogiques en libre accès, cette initiative vise à renforcer l’engagement numérique responsable et à encourager la création de contenu factuel à travers le pays.

Développement

Le Secrétariat de la Zone de libre-échange continentale africaine (ZLECA) a sélectionné le Kenya, le Maroc et le Nigeria comme premiers pays à mettre en œuvre une initiative d’infrastructure publique numérique (DPI) destinée à soutenir le commerce transfrontalier. Cette initiative, baptisée « Africa Digital Access and Public Infrastructure for Trade » (ADAPT), vise à relier l’identité numérique, l’échange de données fiables et les systèmes de paiement interopérables afin de réduire les frictions dans le commerce intra-africain.

L’Australie a mis en place un cadre national pour les normes de santé numérique afin d’améliorer l’interopérabilité et la cohérence entre les systèmes de santé. Ce cadre remédie à la fragmentation causée par des normes de santé numérique développées de manière indépendante. Il fournit également des orientations destinées à soutenir la coordination entre les agences gouvernementales, les prestataires de soins de santé et les acteurs du secteur. Le cadre soutient également l’utilisation de normes terminologiques cliniques reconnues au niveau international et les initiatives de formation associées.

Le Parlement grec a approuvé une loi qui accélère l’interopérabilité entre les systèmes d’information publics, établit une plateforme de services numériques unifiée sur gov.gr, met en place un registre des entités autorisées à utiliser les services gouvernementaux de vérification des données et renforce la cybersécurité grâce à un blocage plus rapide des sites de phishing et à un registre national des domaines malveillants.

Économie

L’OCDE a publié une étude sur le commerce numérique de l’Association des nations de l’Asie du Sud-Est (ASEAN), qui examine la croissance régionale du commerce numérique et les défis réglementaires qui y sont liés. L’OCDE a déclaré que l’ASEAN bénéficie de l’ouverture commerciale, de l’adoption croissante du numérique et de l’évolution des initiatives politiques régionales. Le rapport a noté que la participation inégale et la fragmentation des réglementations nationales pourraient limiter la poursuite de l’intégration du commerce numérique dans la région.

L’étude a identifié des obstacles, notamment des restrictions affectant les flux transfrontaliers de données, les télécommunications, les services numériques et les systèmes de facilitation des échanges. L’OCDE a souligné l’importance de l’harmonisation réglementaire et des progrès vers des systèmes de commerce sans papier. Le rapport aborde également les opportunités liées à l’adoption de l’IA, notamment les réformes concernant les droits de douane, les flux de données et la réglementation des services numériques. Ces conclusions soulignent l’importance de réformes coordonnées pour renforcer le rôle de l’ASEAN dans l’économie numérique mondiale.

Le Pakistan a promulgué la loi de 2026 sur les actifs virtuels (Virtual Assets Act 2026), établissant ainsi un fondement juridique permanent pour l’Autorité pakistanaise de régulation des actifs virtuels (PVARA). Cette mesure officialise la supervision du secteur des actifs numériques en pleine croissance du pays, qui fonctionnait auparavant selon des règles temporaires introduites en 2025. Dans le cadre de ce nouveau dispositif, la PVARA est chargée d’octroyer des licences, de réglementer et de superviser les actifs virtuels et les prestataires de services d’actifs virtuels opérant au Pakistan. La loi confère à l’autorité le pouvoir de délivrer, de suspendre et de révoquer des licences, tandis que les opérations non autorisées s’exposent à des amendes et des sanctions pénales.

La Géorgie s’apprête à lancer le GEL₮, une cryptomonnaie indexé sur le lari géorgien. Selon les responsables, cette initiative vise à soutenir une infrastructure de paiement basée sur la blockchain dans le cadre d’un dispositif réglementaire dédié aux cryptomonnaies. Les responsables ont indiqué que ce projet pourrait favoriser l’accélération des paiements, les transferts transfrontaliers, le développement des technologies financières et les services financiers programmables.La Banque centrale du Brésil a introduit une nouvelle exigence pour les prestataires de services d’actifs virtuels souhaitant obtenir l’autorisation d’opérer dans le pays. À compter du 1er juin 2026, les entreprises souhaitant exercer en tant que « sociedades prestadoras de serviços de ativos virtuais » (SPSAV) devront présenter un rapport d’assurance raisonnable émis par un auditeur indépendant enregistré auprès de l’autorité de régulation des marchés financiers brésilienne, la Comissão de Valores Mobiliários. Cette exigence d’audit vise à évaluer si les candidats disposent de structures de conformité et de contrôle adéquates. Les examens porteront principalement sur les mesures de lutte contre le blanchiment d’argent et le financement du terrorisme, notamment les dispositifs de gouvernance, les procédures de vérification des clients, les contrôles internes des risques et les mécanismes visant à prévenir l’utilisation abusive des services d’actifs virtuels.

Xi reçoit Trump et Poutine en l’espace de deux semaines alors que la diplomatie technologique occupe le devant de la scène

Alors que de nombreux internautes se sont amusés à comparer les visites de Donald Trump et de Vladimir Poutine en Chine jusque dans les moindres détails, y compris les prestations des enfants lors des cérémonies officielles, nous nous concentrerons ici sur les questions technologiques abordées lors de ces rencontres, avec seulement un bref aperçu des similitudes et différences entre les deux séries d’entretiens.

Le sommet Trump-Xi témoigne d’un certain malaise, mais laisse entrevoir de vagues progrès

Le commerce et la technologie figuraient parmi les principaux thèmes censés dominer le sommet Trump-Xi du 13 au 15 mai 2026 à Pékin, la première visite d’État d’un président américain en Chine depuis 8 ans.

Une pléiade de PDG accompagnait Trump lors de ce voyage, notamment Elon Musk de Tesla, Jensen Huang de Nvidia, Tim Cook d’Apple, Larry Culp de GE Aerospace, Kelly Ortberg de Boeing, Dina Powell McCormick de Meta, Larry Fink de BlackRock, Stephen Schwarzman de Blackstone, Sanjay Mehrotra de Micron, Michael Miebach de Mastercard, Cristiano Amon de Qualcomm et Ryan McInerney de Visa.

Mais même si le public est assuré que des progrès ont été réalisés, les détails restent rares.

En matière de technologie, l’IA devait être au cœur des discussions, des informations publiées à l’approche du sommet laissant entendre que les États-Unis et la Chine envisageaient d’entamer des discussions officielles sur l’IA. « Les deux superpuissances de l’IA vont entamer le dialogue. Nous allons établir un protocole sur la manière de mettre en œuvre les meilleures pratiques en matière d’IA afin de nous assurer que des acteurs non étatiques ne mettent pas la main sur ces modèles », a déclaré le secrétaire au Trésor américain Scott Bessent à CNBC.

Les puces nécessaires à l’IA font bien sûr partie intégrante de cette histoire. Les États-Unis ont tenté de freiner le développement de l’IA en Chine en limitant les ventes de puces de pointe, ce qui a pris au piège la société Nvidia de Jensen Huang et Huang lui-même. Pourtant, selon certaines informations, les États-Unis auraient autorisé une dizaine d’entreprises chinoises à acheter la H200, la deuxième puce d’IA la plus puissante de Nvidia. Cependant, aucune livraison n’a été effectuée. C’est probablement la percée que Huang recherchait en Chine.

Une partie de l’histoire des puces concerne les minéraux qui les composent, et sur ce point, la Chine détient un avantage qu’elle utilise habilement comme levier dans ses relations commerciales avec les États-Unis. Le représentant américain au commerce, Jamieson Greer, a déclaré à Bloomberg que les exportations de terres rares en provenance de Chine s’amélioraient, bien que Pékin tarde encore à approuver certaines licences d’exportation. Cependant, M. Greer a également indiqué que les contrôles à l’exportation des puces n’avaient pas constitué un élément majeur des discussions.

Quelques jours plus tard, la Maison Blanche a déclaré que la Chine répondrait aux préoccupations américaines concernant les pénuries dans la chaîne d’approvisionnement liées aux terres rares et à d’autres minéraux critiques, ainsi qu’aux inquiétudes relatives aux restrictions sur la vente d’équipements et de technologies de production et de traitement des terres rares. En réponse aux questions concernant cette déclaration, le ministère chinois du Commerce a indiqué que les deux parties avaient discuté de la question et s’efforceraient de résoudre leurs préoccupations respectives, pour autant qu’elles soient raisonnables et légitimes.

De plus, selon certaines informations, ByteDance aurait accepté d’acheter des millions de puces ASIC de Qualcomm destinées à son infrastructure d’IA et à ses systèmes d’agents en pleine expansion, un accord qui aurait pu être conclu lors de cette visite.

Le sommet Russie-Chine annonce un renforcement de la coopération dans les domaines de l’IA, de l’industrie et de l’économie numérique

Les entretiens entre Poutine et Xi les 19 et 20 mai 2026 ont abouti à un communiqué qui, d’emblée, est révélateur. Le document lui-même est assez long et, bien qu’il ne soit pas très technique, il contient néanmoins des signaux importants.

Poutine et Xi ont souligné que l’IA devait être développée et déployée de manière à servir le développement universel plutôt que des intérêts nationaux étroits. Dans le même temps, ils ont explicitement critiqué l’utilisation de l’IA par certains États comme instrument géopolitique visant à préserver ou à renforcer leur domination mondiale.

Les dirigeants ont également mis en avant la coopération croissante sur la dimension militaire de l’IA. La Russie et la Chine ont l’intention d’approfondir leur collaboration sur le développement et l’application de l’IA dans des contextes militaires, tant par le biais de canaux bilatéraux que dans le cadre de processus multilatéraux, y compris les discussions au sein du Groupe d’experts gouvernementaux sur les systèmes d’armes autonomes létales (LAWS).

Les deux parties se sont engagées à coordonner leurs positions sur les questions scientifiques et politiques liées à l’IA au sein des organisations internationales. Cela témoigne d’un effort visant à harmoniser leurs approches de manière plus systématique dans les forums mondiaux de normalisation et d’élaboration des normes. Il n’est pas inhabituel que la Chine et la Russie adoptent des positions similaires lors de négociations sur les questions numériques, mais il s’agissait là d’un signal géopolitique assez explicite.

Person, Box, Face, Head, Cardboard, Carton, Package, Package Delivery

La Chine et la Russie fusionneront leurs systèmes de navigation par satellite GLONASS et Beidou en un service mondial complémentaire, coopéreront à la coordination et à l’utilisation des fréquences radio et des créneaux orbitaux, et étendront leur coopération en matière d’Internet par satellite et d’Internet des objets.

La coopération en matière de logiciels libres est également apparue comme une priorité stratégique. Les deux pays prévoient d’explorer un mécanisme de coopération bilatérale en matière de logiciels et de promouvoir les technologies libres dans des secteurs clés. Cela s’inscrit dans le cadre d’efforts plus larges menés par les deux pays pour localiser les écosystèmes logiciels et réduire leur dépendance vis-à-vis des plateformes technologiques américaines.

Les parties ont convenu de renforcer leur coopération dans la lutte contre la cybercriminalité, saluant l’adoption par l’Assemblée générale des Nations unies de la Convention des Nations unies contre la cybercriminalité et s’engageant à sa mise en vigueur et à sa mise en œuvre rapides, y compris en travaillant sur un protocole additionnel visant à élargir son champ d’application et à renforcer la coopération internationale.

Elles se sont engagées à approfondir leur coopération stratégique en matière de sécurité de l’information et à coordonner leurs réponses aux menaces pesant sur les TIC. Elles ont également l’intention d’échanger leurs expériences en matière de réglementation législative de l’internet.

Elles ont souligné le rôle central de l’ONU dans la réponse aux menaces dans l’espace informationnel, après quoi les pays ont exprimé leur soutien aux travaux du Mécanisme mondial. Cela n’est pas non plus surprenant, étant donné que le Mécanisme mondial est issu du Groupe de travail à composition ouverte (OEWG), qui était lui-même une initiative russe. Les pays ont également exprimé leur soutien à l’élaboration de documents juridiques internationaux plus larges couvrant des questions telles que la sécurité des données et la résilience de la chaîne d’approvisionnement. Cela s’inscrit dans la lignée des efforts de longue date de la Chine pour mettre en place l’Initiative mondiale sur la sécurité des données.

Les parties ont convenu d’étendre leur coopération dans le domaine de l’économie numérique et du commerce électronique transfrontalier, parallèlement à une collaboration plus large dans des secteurs tels que l’automobile, l’aviation et l’extraction minière. Elles se sont engagées à renforcer la coordination en matière de protection de la propriété intellectuelle et de protection des consommateurs dans les services en ligne et transfrontaliers,

Elles ont également convenu d’approfondir la coopération en matière de commerce électronique par le biais de cadres régionaux, notamment l’Union économique eurasienne (UEE), l’initiative « La Ceinture et la Route », l’Organisation de coopération de Shanghai (OCS), l’APEC et l’Initiative renforcée de Tumangan. La coopération se concentrera sur la numérisation, la logistique, le commerce sans barrières, la connectivité et l’innovation, avec pour objectif plus large de stimuler l’intégration économique régionale, le commerce et l’emploi.

Cette partie n’est pas surprenante, étant donné que la Chine et la Russie ont à plusieurs reprises présenté le commerce électronique, l’économie numérique et la coopération en matière de TIC comme des piliers stratégiques de leur partenariat depuis au moins le milieu des années 2010 dans les contextes de l’OCS et de la « Belt and Road ».

Comparaison des deux sommets

Il n’y a pas eu de communiqué conjoint à l’issue de la rencontre entre Xi et Trump, tandis que le sommet Xi-Poutine a donné lieu à un communiqué détaillé et volumineux. Ce simple fait est assez révélateur. Xi et Poutine ont également signé plus de 20 accords couvrant l’énergie, le commerce, la science et la technologie, ainsi que les infrastructures, alors qu’on ignore encore combien d’accords Xi et Trump ont signés.

Dans le contexte américano-chinois, les commentaires formulés par Bessent lors de la visite de Trump en Chine, soulignant que « la raison pour laquelle nous sommes en mesure d’avoir des discussions constructives avec les Chinois sur l’IA, c’est parce que nous sommes en tête », mettent en évidence la concurrence technologique persistante entre les deux pays. Bessent a également admis sans détour : « Je ne pense pas que nous aurions les mêmes discussions s’ils avaient une telle longueur d’avance sur nous. » Le président Trump a également reporté la signature du décret sur l’IA et la cybersécurité, affirmant qu’il ne souhaitait pas que de nouvelles règles ralentissent le leadership américain en matière d’IA ou affaiblissent son avantage concurrentiel sur la Chine, ce qui renforce le cadre concurrentiel. Cela s’oppose directement à la coopération avec la Russie et à la future coordination sur les questions scientifiques et politiques liées à l’IA au sein des organisations internationales.

L’engagement de la Chine avec la Russie a pris une forme plus coopérative et tournée vers l’avenir. Les deux parties ont convenu de promouvoir activement la coopération en matière d’extraction minière conjointe et d’élaboration de normes écologiques. La Russie détient certaines des plus grandes réserves mondiales de terres rares – elle se classe au cinquième rang mondial – mais reste limitée par des capacités restreintes en matière d’extraction, de raffinage et de traitement, un domaine dans lequel elle a tout à gagner de l’expertise technologique et industrielle chinoise.

Dans l’ensemble, la différence entre les résultats obtenus et l’importance accordée aux enjeux montre que la relation entre la Russie et la Chine semble plus à l’aise pour aborder des domaines sensibles tels que la coopération en matière d’IA et les applications militaires, tandis que l’engagement entre les États-Unis et la Chine reste plus contraint et concurrentiel tant dans le ton que dans la portée.

Le mois dernier à Genève

Geneva Cyber Week 2026

L’Institut des Nations unies pour la recherche sur le désarmement (UNIDIR) et le Département fédéral des affaires étrangères (DFAE) ont co-organisé la Geneva Cyber Week du 4 au 8 mai 2026, réunissant des décideurs politiques, des diplomates, des experts techniques, des dirigeants d’entreprise, des universitaires et des représentants de la société civile dans divers lieux à Genève et en ligne pour une semaine de discussions sur la stabilité cybernétique, la résilience, la gouvernance, la numérisation et les implications sécuritaires des technologies émergentes, y compris l’IA.

De retour après sa première édition, l’événement s’est positionné comme une réponse à un environnement cybernétique et géopolitique plus fragile. Organisée sous le thème « Faire progresser la coopération mondiale dans le cyberespace », la Geneva Cyber Week 2026 s’est déroulée dans un contexte d’insécurité cybernétique croissante, de tensions géopolitiques s’intensifiant et d’évolution technologique rapide. Le programme comprenait près de 90 événements et a renforcé le rôle de Genève en tant que centre de cyberdiplomatie, de coopération internationale et de gouvernance numérique.

Dans le cadre de la Geneva Cyber Week, l’UNIDIR a organisé la Conférence sur la stabilité cybernétique 2026, les 4 et 5 mai à Genève et en ligne, réunissant des représentants des gouvernements, des organisations internationales, du secteur privé, du monde universitaire et de la société civile pour débattre de la sécurité des TIC et de la gouvernance cybernétique. Sous le thème « La cybergouvernance à l’ère de la révolution technologique : leçons du passé, réalités du présent et frontières de l’avenir », les discussions ont porté sur la manière dont les cadres internationaux de cyberstabilité s’adaptent à l’évolution technologique rapide, notamment l’IA et l’informatique quantique, tout en tirant les leçons des processus de cyberdiplomatie passés et des défis actuels en matière de sécurité.

Réunion d’experts pluriannuelle sur l’investissement, l’innovation et l’entrepreneuriat pour le renforcement des capacités productives et le développement durable, 12e session

La réunion d’experts pluriannuelle de la CNUCED sur l’investissement, l’innovation et l’entrepreneuriat pour le renforcement des capacités productives et le développement durable s’est réunie pour sa douzième session les 4 et 5 mai. Les experts ont averti que l’IA et d’autres technologies stratégiques sont en train de remodeler les schémas d’investissement mondiaux, concentrant les capitaux dans une poignée de secteurs et de pays tout en laissant de nombreuses économies en développement à la traîne. Les discussions lors de la réunion ont porté sur la manière dont les pays en développement peuvent être compétitifs dans les secteurs liés à l’IA, renforcer leurs écosystèmes d’innovation nationaux et veiller à ce que les investissements axés sur l’IA se traduisent par des gains de développement plus larges.

Ouverture à Genève de la conférence de l’OSCE intitulée « Anticiper les technologies – pour un avenir sûr et humain »

La présidence suisse de l’Organisation pour la sécurité et la coopération en Europe a ouvert le 7 mai à Genève une conférence de haut niveau de deux jours sur les technologies anticipatives. L’événement a examiné comment la prospective, le dialogue et la coopération internationale peuvent contribuer à réduire les malentendus, à instaurer la confiance et à renforcer la sécurité dans toute la région de l’OSCE dans un contexte d’évolution technologique rapide.

Le programme comprenait des discussions sur l’anticipation des changements technologiques et leur impact géopolitique, la sécurité de l’eau et de l’énergie à l’ère numérique, ainsi que le rôle de l’IA dans l’alerte précoce et la prévention des conflits.

La conférence a également mis en avant le rôle de Genève en tant que point de rencontre entre science et diplomatie, notamment à travers des institutions telles que le CERN, le Geneva Science and Diplomacy Anticipator et l’Open Quantum Institute.

Cet événement s’inscrit dans le cadre de la priorité de la présidence visant à relier l’anticipation scientifique et technologique à l’action politique.

L’OMC a repris les négociations alors que 19 membres soutiennent l’engagement en faveur d’un moratoire sur le commerce électronique

Le Conseil général de l’OMC s’est réuni à Genève le 8 mai pour la première fois depuis la 14e Conférence ministérielle (CM14), après que les négociateurs à Yaoundé ont manqué de peu de parvenir à des accords sur plusieurs dossiers majeurs, notamment l’avenir du moratoire de longue date sur le commerce électronique et la réforme plus large de l’OMC.

La nouvelle présidente, l’ambassadrice Clare Kelly, a déclaré que les membres restaient déterminés à préserver le fragile équilibre atteint lors des négociations au Cameroun et à éviter un retour aux positions antérieures.

Ces discussions font suite à l’expiration, le 31 mars, du moratoire de l’OMC sur les droits de douane applicables aux transmissions électroniques, un accord temporaire adopté pour la première fois en 1998 qui empêchait les pays d’imposer des droits de douane sur les flux commerciaux numériques tels que les téléchargements de logiciels, les services de streaming et autres transmissions en ligne. Les ministres présents à la MC14 ne sont pas parvenus à s’entendre sur une nouvelle prolongation, révélant de profondes divisions quant à la durée de maintien du moratoire et à la manière dont les gouvernements devraient réagir aux évolutions technologiques rapides telles que l’IA et l’impression 3D.

Malgré l’expiration du moratoire, un groupe de 19 membres de l’OMC a annoncé qu’il continuerait à ne pas imposer de droits de douane sur les transmissions électroniques entre eux. Dans une déclaration commune diffusée à l’OMC, le groupe a déclaré que cet accord apporterait prévisibilité et certitude aux entreprises et aux consommateurs pendant que les négociations multilatérales se poursuivent. Le groupe comprend l’Argentine, l’Australie, le Costa Rica, l’Équateur, le Guatemala, l’Islande, Israël, le Japon, la Corée du Sud, la Malaisie, le Mexique, la Nouvelle-Zélande, la Norvège, le Panama, le Paraguay, Singapour, le territoire douanier distinct de Taïwan, Penghu, Kinmen et Matsu, les États-Unis et l’Uruguay.

Türkiye a également fait preuve d’une nouvelle souplesse lors de la réunion du Conseil général, en annonçant qu’elle ne ferait pas obstacle à un consensus sur une prolongation temporaire du moratoire. Le Brésil a toutefois maintenu son opposition à une prolongation de quatre ans du moratoire.

Le président a annoncé de nouvelles consultations sur le commerce électronique et la réforme de l’OMC, avec l’intention de faire rapport aux membres en juillet.

Groupe intergouvernemental d’experts sur le commerce électronique et l’économie numérique – 9e session

Le Groupe intergouvernemental d’experts de la CNUCED sur le commerce électronique et l’économie numérique s’est réuni à Genève pour sa 9e session du 11 au 13 mai 2026 afin d’examiner l’impact du commerce numérique sur les systèmes fiscaux des pays en développement. L’ordre du jour était axé sur le renforcement des recettes fiscales dans le contexte du commerce électronique, notamment les défis liés à la TVA et aux douanes, la taxation des plateformes et les implications des transactions numériques transfrontalières. Les sessions ont porté sur les risques liés aux recettes, les outils numériques administratifs et le rôle des plateformes en tant qu’intermédiaires fiscaux, ainsi que sur la coopération internationale et régionale visant à améliorer la préparation fiscale. La réunion a également abordé la mesure de l’économie numérique et le renforcement des capacités pour soutenir des stratégies de développement numérique inclusives et durables.

Journée des mondes virtuels des Nations Unies 2026

La 3e Journée des mondes virtuels des Nations unies, co-organisée par l’Union internationale des télécommunications (UIT) et plusieurs agences et partenaires des Nations unies, s’est tenue à Genève les 11 et 12 mai 2026. Elle a réuni des ambassadeurs, des ministres, des dirigeants municipaux, des entités des Nations unies, des pionniers du secteur et des innovateurs pour deux jours de dialogue de haut niveau sur les villes pilotées par l’IA, les environnements immersifs et l’intelligence spatiale. Les discussions ont porté sur la manière dont les technologies de pointe redéfinissent la gouvernance urbaine et les services publics, notamment le concept émergent de « citiverse » et son lien avec le Pacte numérique mondial. L’événement s’est conclu par une réunion d’information à l’intention des dirigeants sur l’IA, l’intelligence spatiale et le « citiverse » basé sur l’IA, ainsi que par un appel à l’action pour un avenir numérique inclusif.

79e Assemblée mondiale de la santé

La 79e Assemblée mondiale de la santé (AMS79), qui s’est tenue à Genève du 18 au 23 mai 2026, a réuni les États membres de l’OMS, des organisations internationales, la société civile et les acteurs du secteur de la santé pour discuter des priorités et de la gouvernance en matière de santé mondiale. Parallèlement aux travaux officiels de l’Assemblée, un large éventail d’événements parallèles a été consacré à la santé numérique et à l’IA, abordant des thèmes tels que la fiabilité des données de santé, l’interopérabilité, la gouvernance responsable de l’IA et la transformation numérique des systèmes de santé. Les discussions ont examiné comment l’IA et les technologies numériques peuvent soutenir la couverture sanitaire universelle, renforcer l’élaboration de politiques fondées sur des données probantes, améliorer les diagnostics et les soins aux personnes âgées, et renforcer la coopération transfrontalière, tout en abordant les questions d’éthique, de transparence, de gouvernance des données et d’équité dans les systèmes de santé.

Les experts des droits de l’Homme de l’ONU s’inquiètent de l’exploitation sexuelle à grande échelle des femmes et des filles en ligne

La Rapporteuse spéciale des Nations unies sur la violence à l’égard des femmes et des filles, ses causes et ses conséquences, et le Rapporteur spécial des Nations unies sur le droit à la vie privée ont publié une déclaration exprimant leurs préoccupations face à ce qu’ils décrivent comme une exploitation sexuelle à grande échelle des femmes et des filles, facilitée et monétisée par Pornhub et sa société mère, Aylo Holdings, ainsi que par les réseaux de paiement et les moteurs de recherche qui permettent la distribution et la monétisation de ce contenu.

Les rapporteurs se sont concentrés sur Pornhub et sa société mère Aylo Holdings, tout en évoquant des préoccupations plus générales concernant les plateformes de pornographie générée par les utilisateurs, les réseaux de paiement et les entreprises de recherche et de technologie liées aux systèmes de distribution en ligne. Les experts ont déclaré que les entreprises impliquées dans les écosystèmes de contenu numérique ne devraient pas se soustraire à leurs responsabilités lorsque leurs services contribuent à des violations des droits de l’homme. Ils ont appelé à la mise en place de garanties plus solides, notamment des systèmes obligatoires de vérification de l’âge et du consentement par des tiers pour les plateformes de pornographie générée par les utilisateurs.

Le Haut-Commissariat des Nations Unies aux droits de l’Homme publie des lignes directrices sur la sécurité des enfants en ligne

Une nouvelle note d’information du Haut-Commissariat des Nations Unies aux droits de l’Homme (HCDH) présente dix mesures interdépendantes visant à améliorer la sécurité des enfants en ligne grâce à une approche réglementaire fondée sur les droits et s’inscrivant au niveau du système. Plutôt que de s’appuyer sur des restrictions générales ou de rejeter la responsabilité sur les utilisateurs, elle donne la priorité à des changements structurels dans la conception et la gouvernance des plateformes.

Tout d’abord, il préconise des obligations de « sécurité dès la conception » visant à réduire les fonctionnalités préjudiciables telles que la lecture automatique, le défilement infini et les notifications intrusives. La réglementation doit être fondée sur les droits des enfants, en veillant à ce que la protection soit équilibrée avec les droits à la participation, à l’expression, à la vie privée et à l’accès à l’information. Des mesures solides de protection de la confidentialité des données sont essentielles, notamment l’interdiction du micro-ciblage et la limitation du profilage des enfants à des fins commerciales.

Les entreprises sont tenues de mener des évaluations continues de l’impact sur les droits de l’enfant dans le cadre d’une diligence raisonnable plus large en matière de droits humains, en faisant preuve de transparence et en rendant compte de leurs conclusions. Lorsque la vérification de l’âge est utilisée, elle doit s’accompagner de garanties strictes concernant la collecte et le stockage des données, ainsi que la non-discrimination.

Les lignes directrices mettent l’accent sur des restrictions d’âge ciblées et spécifiques aux préjudices plutôt que sur des interdictions générales, et soulignent que les enfants devraient être activement impliqués dans l’influence des politiques qui affectent leurs environnements numériques. Les plateformes devraient également être tenues d’assurer une transparence significative sur les systèmes de recommandation, les pratiques de modération et les choix de conception.

Enfin, une surveillance efficace, des mécanismes de responsabilisation contraignants et un processus décisionnel continu fondé sur des données factuelles sont essentiels pour garantir que la réglementation s’adapte à l’évolution des technologies et des risques.

Il convient de noter que ces lignes directrices considèrent la sécurité des enfants en ligne comme une question à la fois de gouvernance et de droits. Elles soutiennent que les droits des enfants s’appliquent en ligne au même titre qu’hors ligne, y compris les droits liés à la participation, à la vie privée, à l’accès à l’information et à la liberté d’expression. Les enfants, précise le document, devraient également avoir leur mot à dire sur la manière dont les espaces numériques sont gouvernés et réglementés.

À ne pas manquer !

Suivez la préparation du Sommet de l’IA de Genève 2027, une initiative mondiale destinée à façonner l’avenir de l’intelligence artificielle. Ajoutez cette page à vos favoris et inscrivez-vous aux mises à jour afin de rester informé et engagé tout au long du parcours menant à Genève 2027.

Cela vaut-il la peine d’approfondir ?

Weekly #265 ‘Tech Liberation Day’: EU launches tech sovereignty push

29 May – 4 June 2026

HIGHLIGHT OF THE WEEK

‘Tech Liberation Day’: EU launches tech sovereignty push

The European Commission has presented its long-awaited European Technological Sovereignty Package to strengthen Europe’s capacity in semiconductors, AI, cloud infrastructure, and open-source technologies. We examine what tech sovereignty is, what it is not, the building blocks of the package, and the context behind it.

IN OTHER NEWS LAST WEEK

AI governance

The UN. The UN Office for Digital and Emerging Technologies (ODET) has launched the AI Governance for Humanity Lab in Valencia to strengthen international cooperation on AI governance. The Lab will focus on improving interoperability between national and regional governance frameworks and supporting practical implementation across regions and sectors. Its work will include network mobilisation, comparative policy analysis, and the development of cooperative tools for AI governance.

G7. The G7 Digital and Technology Ministers agreed on a set of priorities covering secure AI, AI openness, digital sector resilience and online safety for minors following a meeting in Paris under France’s presidency. The ministers committed to advancing secure, responsible, and trustworthy AI by continuing to implement the Hiroshima AI Process and its reporting framework, while broadening international participation and cooperation in AI risk assessment and governance. They plan to launch discussions to improve the comparability of AI evaluation frameworks, promote the adoption of AI procurement guidelines in the public sector, and strengthen collaboration on detecting AI-generated content, including through the development of synthetic-content detection tools. To boost AI adoption, they called for clearer principles on AI openness, greater support for AI literacy and skills development, and targeted measures to help SMEs adopt AI, including a new G7 SME AI Readiness Tool and expanded access to training resources.

The OECD. The Organisation for Economic Co-operation and Development (OECD) has launched the AI Policy Toolkit, a practical guide intended to help governments translate AI principles into policy action. The toolkit is intended to help governments identify policy priorities, compare international approaches and adapt guidance to national circumstances. The platform incorporates AI-powered semantic search to help users identify relevant policy examples and practical approaches drawn from real-world experience.

The EU. The European Commission has appointed two new expert bodies to support the implementation and enforcement of the EU’s AI Act. The Scientific Panel and Advisory Forum will provide independent expertise to the Commission’s AI Office and national authorities responsible for supervising compliance with the regulation. The panel will focus on general-purpose AI (GPAI) models and systems, systemic risk assessment, model classification, evaluation methodologies, and cross-border market surveillance. The forum will bring together representatives from academia, civil society, industry, startups and SMEs and provide technical advice on implementation challenges, standardisation efforts, and broader issues related to the enforcement of the AI Act.

The USA. US President Donald Trump has issued an executive order aimed at strengthening US cybersecurity through closer cooperation with the AI industry while preserving a light-touch approach to AI regulation to maintain American technological leadership. The order directs federal agencies to accelerate the protection of government networks, expand access to AI-enabled cybersecurity tools, and improve support for critical infrastructure operators, including hospitals, banks, and utilities. The order also establishes a voluntary AI cybersecurity clearinghouse to coordinate vulnerability discovery and patching efforts across government and industry. In addition, federal authorities will develop a classified process for assessing advanced AI systems and identifying “covered frontier models” with significant cyber capabilities. Developers will be encouraged, but not required, to provide the government with early access to such models before wider release.

Canada. Canada has launched AI for All, a national strategy to accelerate AI adoption while strengthening digital sovereignty and positioning the country as a leading AI economy. The strategy rests on three pillars: trust, opportunity, and sovereignty. On trust, Canada plans to modernise digital legislation, strengthen data protection, address risks such as deepfakes and surveillance pricing, introduce online safety rules, and expand the Canadian AI Safety Institute. On opportunity, it will roll out a national AI literacy initiative, support SMEs and worker reskilling, provide access to AI tools for students, and launch sectoral “AI Missions,” including a flagship health programme focused on diagnostics and system efficiency.

On sovereignty, the plan prioritises domestic AI infrastructure—covering compute, cloud, connectivity, data, and talent—through investments in a national AI supercomputer, sovereign cloud capacity, improved financing for AI firms, strategic procurement, and expanded talent programmes.

Australia. Australia has launched an AI safety institute, aiming to analyse and test new AI models and applications, support regulators and agencies in responding to emerging AI harms, and shape safe AI development, deployment and international governance in Australia’s interests.

The UK. The UK’s signals intelligence agency GCHQ has announced plans to develop an AI-powered national cyber defence capability that would use autonomous software agents to identify and respond to cyber threats at machine speed. Speaking publicly, GCHQ director Anne Keast-Butler described the initiative as a ‘blueprint for a new national cyber defence capability’ to be operational within five years. The programme would apply agentic AI to monitor and protect critical sectors, including energy, water, healthcare, transport, and financial services.

India-South Africa. India and South Africa have agreed to strengthen bilateral cooperation in emerging technologies, with AI, digital infrastructure and advanced manufacturing identified as key areas for future collaboration. Discussions covered biotechnology, genomics, vaccine development, health technologies, renewable energy, hydrogen, advanced manufacturing and digital innovation. The two countries also explored opportunities to deepen cooperation in quantum technologies, geospatial technologies and digital infrastructure.

The UK-France. The UK and France have launched a science and technology partnership focused on applying AI, advanced imaging, and data science to major healthcare challenges, including women’s health, infectious diseases, and antimicrobial resistance.

Cybersecurity: Attack and defend

Germany’s federal cabinet has approved draft legislation that would expand cyber defence capabilities for three federal agencies, the Federal Office for Information Security (BSI), the Federal Criminal Police Office (BKA), and the Federal Police (Bundespolizei). Authorities would be able to block or disrupt software and server infrastructure used in cyberattacks, including systems located outside Germany. The BSI would also receive expanded authority to collect, store, and analyse data to detect activities indicative of attack preparation.

This reflects a broader international trend where cyber operations are becoming more proactive, but defence is only one side of the coin.

In parallel, reporting suggests that the USA is exploring the use of advanced AI systems in offensive cyber operations. According to media reports, the US National Security Agency (NSA) may be using Anthropic’s Claude Mythos model to support cyber operations, with Anthropic reportedly deploying Forward Deployed Engineers to assist integration and deployment. The model is described as highly capable in vulnerability discovery and cyber exploitation scenarios, raising questions about the dual-use nature of frontier AI systems and their role in both defence and offensive planning.

While Germany and the USA focus on hardening state-level cyber power, efforts elsewhere are persisting in defending vulnerable users: children. Malaysia started banning children under 16 from creating social media accounts on 1 June. Major platforms such as Facebook, Instagram, TikTok, and YouTube have been required to verify users’ ages against government records, with penalties of up to 10 million ringgit (around US$2.5 million) for non-compliance. Existing users will be subject to age verification checks during a 6-month transition period.

Meanwhile, Sweden is also considering a social media ban for young users. In an interim report submitted to the Minister of Social Affairs, government investigators appointed last year have recommended that under-15s be banned from platforms that enable users to discover, connect with, and communicate with each other, or where content can be shared or discovered.

Both moves reflect a growing consensus that unmediated access to social media carries risks for younger populations, and that age verification is becoming a regulatory baseline.

That consensus finds its most formal expression at the multilateral level. The G7 Digital and Technology Ministers agreed on a Common Set of Principles defining a safer and more secure digital space for minors. The ministers emphasise effective age assurance, safety-by-design platform features, including default protections, parental controls, and recommendation systems that limit exposure to harmful or addictive content. They also call for strong prevention and enforcement against severe harms such as child sexual abuse material and non-consensual intimate imagery, including AI-generated forms, alongside better support for victims. In parallel, they emphasise empowering parents with usable control tools, strengthening digital and AI literacy for minors, improving transparency and user control over data, and ensuring platforms systematically assess, mitigate, and report risks. The principles also stress coordinated action between governments, industry, researchers, educators, and civil society.

China updates trade secret protection rules to include digital assets

China has updated its trade secret protection rules to explicitly include digital assets such as data, algorithms, software code, and computer programs. The new framework, in force since 1 June 2026, modernises rules from 1995 and strengthens requirements for corporate safeguards, including confidentiality agreements, employee training, access controls, encryption, and data management practices. It also introduces detailed provisions for digital and remote work environments, encouraging tools such as permission-based access, data masking, and activity logging, while explicitly banning illicit methods, such as hacking, fraud, and unauthorised system access to obtain trade secrets. The rules further extend protection to overseas infringements that affect China’s domestic market or the rights of Chinese companies.

Why does it matter? It gives Chinese companies a stronger legal shield to protect AI algorithms, software code, and data from foreign espionage or theft, especially amid US-China tech decoupling. It also signals Beijing’s intent to enforce economic security laws extraterritorially, potentially triggering cross-border legal disputes and raising compliance risks for multinational firms operating in or with China.

France fines Shein over consumer protection breaches

France’s consumer watchdog has imposed two administrative fines on companies linked to Shein after finding consumer protection and environmental disclosure breaches on the retailer’s French website.

The Directorate General for Competition, Consumer Affairs and Fraud Control said an investigation carried out in 2025 on fr.shein.com found failures linked to the right of withdrawal, environmental product information, and order confirmation requirements.

Infinite Styles Ecommerce Co Limited, the seller of Shein-branded products on the French site, was fined €5.76 million. The investigation found that consumers were unable to cancel purchases under the legally required withdrawal procedures. It also found missing information on product traceability and the presence of plastic microfibres in certain textile products. The watchdog said consumers must be informed when textiles containing more than 50% synthetic fibres release plastic microfibres into the environment during washing.

A second company, Infinite Styles Services Co Limited, which operates fr.shein.com, was fined €16.73 million for non-compliant order confirmations. The DGCCRF said confirmations sent to consumers were missing mandatory information, including the price of goods, delivery dates or deadlines, seller identity and contact details, legal guarantees, mediation options, and withdrawal forms and rights. French authorities said the missing information weakened consumer protection by making it harder for customers to exercise rights such as cancelling purchases or seeking refunds.

Reminder. In February, France announced it would step up its pushback against ultra-low-cost online retailers, with Minister for Small and Medium Enterprises, Trade, Crafts, Tourism, and Purchasing Power Serge Papin declaring 2026 a ‘year of resistance’ to platforms such as Shein. The government argues that physical French shops face strict rules and liability, while global online marketplaces operate under looser standards, creating unfair competition. The latest fines are in line with this policy.

Russia orders safeguards for essential services during mobile internet shutdowns

The move sits within a broader pattern of intermittent mobile internet disruptions, expanding app bans, and tighter controls on VPN use and foreign platforms. Previous reporting has documented how connectivity disruptions in major cities, including Moscow, have led residents to rely on offline or low-tech alternatives such as paper maps, walkie-talkies, and older messaging devices to maintain coordination during outages.

Prime Minister Mikhail Mishustin and FSB Director Alexander Bortnikov are to report back on their progress by 1 July.

View all updates

LAST WEEK IN GENEVA

The WTO organised the Trade and Environment Week 2026 from 1–5 June 2026 at the WTO Headquarters in Geneva and online. The 2026 event included sessions on topics such as eco-design and digital product passports, carbon accounting and emissions standards, trade in environmental goods and services, biodiversity, climate-related border measures, fossil fuel subsidy reform, and the role of trade in supporting global climate action.

The 114th International Labour Conference convened in Geneva from 1–12 June 2026, bringing together governments, employers, and workers from ILO member states. The agenda included standard-setting discussions on decent work in the platform economy, a general discussion on advancing gender equality in the world of work, and oversight of the implementation of international labour standards. Delegates also considered programme, budget, and institutional matters, including supervisory reports and procedural questions.

Geneva in Action: Scaling School Connectivity Worldwide (Part II) was held on 3 June at the Giga Connectivity Centre in Geneva. Co-hosted with Diplo, the 75-minute session brought together Geneva-based organisations and government officials participating in the Giga Learning Hub. The discussion focused on scaling school connectivity within broader digital transformation and e-government reforms, with emphasis on national ownership, global frameworks, and the balance between digital sovereignty and locally adapted technology solutions.

READING CORNER

Beyond the LLM hype: Why the real AI breakthrough is deterministic systems – Diplo

Why do advanced LLMs fail at basic math and logic? Explore an engineering perspective on the return of deterministic AI and why the future of digital governance and industry lies in hybrid systems that combine symbolic reasoning with natural language interfaces.

DON’T MISS

Follow the road to the 2027 Geneva AI Summit, a global journey shaping AI’s future. Bookmark the page and register for updates to stay informed and involved every step of the way to Geneva 2027.

Digital Watch newsletter – Issue 110 – May 2026

May 2026 in retrospect

Dear readers,

Pope Leo XIV has issued Magnifica Humanitas, a landmark encyclical setting out the Catholic Church’s position on AI. The document directly challenges the technocratic model of AI governance, warns against the concentration of power in a small number of technology companies, and questions transhumanist trajectories promoted by leading industry voices. We take a closer look at its key arguments.

Our snapshot of broader headlines goes beyond AI, with legal and cybersecurity developments continuing to dominate the agenda.

We also examine two high-level diplomatic summits held within weeks of each other, where China’s Xi Jinping hosted US President Donald Trump and later Russian President Vladimir Putin. Our focus is on the technology dimensions of these engagements, with a brief comparative look at how the two rounds of talks intersect and diverge.

Finally, we turn to Geneva, which saw a particularly active May across the digital and tech policy space. This edition provides a concise recap of key events and emerging developments from the city’s packed agenda.

AI governance in May

Magnifica humanitas: Human dignity in the age of AI

Pope Leo XIV has issued Magnifica Humanitas, a groundbreaking encyclical outlining the Catholic Church’s approach to AI.

To understand its significance, it is worth noting that an encyclical pastoral letter is a circular letter issued by the pope to the whole Roman Catholic Church on matters of doctrine, morals, or discipline. It is also aimed beyond the Church, often shaping global ethical and policy debates on major issues like inequality, war, the environment, and technology.

When critics objected to Pope Leo XIII publishing Rerum Novarum on society, the economy and politics, the Pope noted that the proclamation of the Gospel cannot overlook the concrete lives of people. This is the same logic behind Magnifica Humanitas.

Anchored in millennia of history, Magnifica Humanitas provides an important antidote to the AI discourse framed as inevitability: the race cannot be stopped, the market cannot be slowed, geopolitical competition cannot be escaped, and humans must simply follow.

The encyclical rejects this inevitability. It insists that technological futures should be shaped by our choices.

The encyclical frames AI governance through two contrasting metaphors: the Tower of Babel and the walls of New Jerusalem. Babel represents technological ambition without humility, reaching upward through ambition, control, and self-exaltation, only to produce fragmentation, confusion, and collapse. The Jerusalem wall metaphor offers a different vision. It is built piece by piece, through cooperation, solidarity, dialogue, and responsibility. Furthermore, it requires diplomacy rather than domination, patience rather than acceleration, and the common good rather than the race for supremacy.

Ultimately, Magnifica Humanitas can be boiled down to a choice between two metaphors: Babel and Jerusalem. The encyclical warns us about sleepwalking into Babel and offers guidance to reaching Jerusalem by drawing on the cognitive, emotional, and spiritual strengths that define us as human beings.

How to fix the ‘operating system’ of humanity. The encyclical frames humanity’s “operating system” as rooted in Axial Age traditions that defined human moral, rational, and spiritual identity, later refined through Enlightenment-era thought. It warns that transhumanism and posthumanism, driven by AI and bioengineering, risk recasting humans as entities to be optimised rather than inherently dignified beings. Magnifica Humanitas focuses on this key dilemma, removing all trappings of techno, business, and other narratives that dominate our communication space. It also calls to revisit the core of humanity’s ‘operating system’ centred around the life and dignity of humans, arguing that humans should not be ‘optimised’ in the same way as technologies.

The encyclical warns about the danger of regarding AI systems as neutral and objective, noting that technology takes on the characteristics of those who devise, finance, regulate and use it.

Clarity on AI governance and accountability. On AI governance, the encyclical offers clarity. It challenges the technocratic paradigm, in which governance is treated as an expert exercise focused mainly on standards, risk management, and efficiency. Instead, the encyclical argues that AI governance must begin with human dignity and the common good. Decisions about AI should be made by the people and communities directly affected. Communities should not merely be consulted after rules and platforms have been set elsewhere. They must be able to shape, challenge, and correct the systems that impact their lives. This resonates strongly with the multistakeholder tradition in internet governance, and challenges both corporate monopoly and state-centric control.

Fresh look at human rights. The encyclical points to a paradox of our age: human rights are widely formally proclaimed, yet rapidly eroded to, among others, by technological progress. Through profiling, manipulating and overall technological optimising, people are increasingly treated as data points and parts of algorithmic patterns. Therefore, one of the encyclical’s most critical contributions is its renewed grounding of human rights to protect our core humanity in the AI era.

Disarming AI. A striking concept in the encyclical is the call to ‘disarm AI.’ This is not limited to the regulation of autonomous weapons or military systems, although those are urgent concerns. It is a wider call for cognitive, economic, and political disarmament. AI is becoming an instrument in geopolitical rivalry, commercial dominance, surveillance, propaganda, and social control. The encyclical challenges the assumption that the AI race is inevitable. It calls for AI to be developed as an enabling tool for humanity rather than as a weapon of power and war.

Slavery and colonialism in the AI era. Digital colonialism is another concern of the encyclical. Colonial powers no longer need to control territory. They can appropriate data, shape markets, control infrastructures, and extract value from human lives transformed into exploitable information. A genuinely post-colonial digital order would return agency to individuals and communities. It would give people not only access to their data but meaningful control over how that data is used, by whom, and for whose benefit.

How to deal with AI risks. Magnifica Humanitas provides an informed and impactful survey of AI risks. On existential risk, it departs from narratives centred on AGI or super-intelligence takeover, arguing instead that existential risks may emerge more gradually, and not necessarily from technology itself, but from those who own and operate AI systems. On exclusion risk, it warns against concentration of power in the hands of a few and exclusion of many, with monopolies taking epistemic, economic, and political forms. It highlights how opacity and lack of public oversight can produce dependencies, exclusions, manipulations, and inequalities, including through control over visibility, amplification, and behavioural influence. It also examines existing risks in jobs, the economy, disinformation, and education.

Language and culture in the AI era. The Pope warns that violence often begins with language: wars are preceded by wars of words. Here, AI plays the critical role in shaping culture as a generator of text, video and other communication artefacts. The encyclical, therefore, calls for renewed attention to truth, dialogue, and the moral ecology of language. AI poses the main risks of generating synthetic narratives, shallow persuasion, and automated distortion. The challenge is not only misinformation. It is the gradual erosion of shared reality.

Education in the age of immediacy. Magnifica Humanitas places great emphasis on education in the AI era, in line with the Catholic Church’s long teaching tradition. The encyclical identifies a key tension between the rapid access to information in the digital age and the inherently slow nature of learning. Information can be delivered instantly. Knowledge and even more wisdom require a lot of time. However, schools and universities are not prepared to address the tension between AI technology and human learning.

Impact of AI on nature and the environment. The encyclical also reminds us that nothing in the digital world is immaterial or magical. AI depends on physical infrastructures: data centres, energy grids, minerals, water, devices, cables, and supply chains. Current AI systems require enormous amounts of energy and water and place growing pressure on natural resources. The environmental cost of AI is therefore not external to the debate. A technology that claims to advance humanity cannot be evaluated only by speed, accuracy, or profitability. It must also be judged by its impact on the planet, on vulnerable communities, and on future generations.

Why Magnifica Humanitas matters. The historic importance of Magnifica Humanitas lies in its attempt to shift the AI debate from technical management to civilisational reflection.

It challenges the concentration of power in a few companies. It questions transhumanist and posthumanist narratives that treat the human person as obsolete or upgradeable. It insists that AI governance must be rooted in dignity, subsidiarity, solidarity, justice, and the common good. It provides new perspectives on the risks of digital slavery, colonialism, militarisation, manipulation, and anthropological regression.

Most importantly, it restores moral agency. Metaphorically speaking, we are not condemned to build a Tower of Babel. We can choose to build the walls of the new Jerusalem.

This text is an adaptation of Dr Jovan Kurbalija’s analysis of Magnifica Humanitas, which will be expanded in the coming days. To receive in-depth analysis, you can register here.

Beyond AI: The developments that made waves in May

Technologies

Norway joined the Pax Silica initiative, which focuses on secure AI, semiconductor, and critical raw materials supply chains. Norwegian officials said participation could improve market access for domestic companies operating in advanced technological sectors and strengthen economic security cooperation with strategic partners.

The European Commission adopted a proposal for the future allocation of the EU’s 2 GHz mobile satellite services spectrum band beyond 2027. The spectrum is increasingly viewed as a strategic asset for both commercial connectivity and defence-related communications, particularly for direct-to-device (D2D) satellite services and critical communications infrastructure. Under the proposal, one-third of the spectrum would be reserved for governmental, security, and military use, tied to the EU’s IRIS² secure connectivity programme. The remaining two-thirds would support commercial services, but half of that capacity would be reserved specifically for EU operators entering the market, while the rest would remain open to both EU and non-EU companies.

Cybersecurity

The EU has adopted common templates for cybersecurity incident reporting across the bloc to reduce administrative burdens and simplify compliance for companies required to report cybersecurity incidents under NIS2. The Commission now plans to adopt the templates through an implementing act, which would make them mandatory for all member states.

The Council of the European Union has extended restrictive measures against individuals and entities involved in cyber-attacks threatening the EU and its member states until 18 May 2027. The framework allows the EU to impose targeted sanctions on persons or entities involved in significant cyber-attacks that constitute an external threat to the Union or its member states. Measures can also be imposed in response to cyber-attacks against third countries or international organisations, where they support Common Foreign and Security Policy objectives.

YouTube, Snap, TikTok, and Meta have agreed to settle a landmark lawsuit brought by Kentucky’s Breathitt County School District, which accused them of designing algorithmic systems and features—such as infinite scrolling and engagement-driven recommendation loops—that foster compulsive use among young users and place additional financial and educational burdens on schools. The terms of the settlement have not been disclosed. The case was scheduled to go to trial on 12 June, which will now not proceed.

In parallel, the UK’s Ofcom has reported that almost all 8-17-year-olds in the UK are online (99%). reinforcing its push for stronger platform accountability. It wants child safety to be embedded into product design from the outset, with regulators informed in advance of new features to assess potential risks. However, the regulator remains concerned that some platforms are not doing enough to make feeds safer and is sceptical that age-13 minimum policies effectively prevent underage access.

The Office of the Privacy Commissioner of Canada has issued guidance on how organisations should assess and implement age assurance tools for websites and online services. The OPC states that age assurance should only be used where there is a clear legal requirement or a demonstrable risk of harm to children. It emphasises that organisations must evaluate whether alternative, less intrusive measures could address these risks before adopting such systems.

Malaysia’s Communications and Multimedia Commission (MCMC) has published the Child Protection Code and Risk Mitigation Code under the Online Safety Act 2025. The new rules require stronger safeguards for young users on online platforms, including stricter age-related controls and enhanced content governance obligations for service providers. Malaysia also started banning children under 16 from creating social media accounts. Major platforms such as Facebook, Instagram, TikTok, and YouTube are required to verify users’ ages against government records, with penalties of up to 10 million ringgit (around US$2.5 million) for non-compliance. Existing users will be subject to age verification checks during a 6-month transition period.

Meanwhile, Sweden is also considering a social media ban for young users. In an interim report submitted to the Minister of Social Affairs, government investigators appointed last year have recommended that under-15s should be banned from platforms that enable users to discover, connect with, and communicate with each other, or where content can be shared or discovered.

The G7 Digital and Technology Ministers agreed on a Common Set of Principles defining a safer and more secure digital space for minors. The ministers emphasise effective age assurance, safety-by-design platform features, including default protections, parental controls, and recommendation systems that limit exposure to harmful or addictive content. They also call for strong prevention and enforcement against severe harms such as child sexual abuse material and non-consensual intimate imagery, including AI-generated forms, alongside better support for victims. In parallel, they emphasise empowering parents with usable control tools, strengthening digital and AI literacy for minors, improving transparency and user control over data, and ensuring platforms systematically assess, mitigate, and report risks. The principles also stress coordinated action between governments, industry, researchers, educators, and civil society.

Human rights

As of 8 May, end-to-end encrypted messaging on Instagram is officially over. Meta has switched off the feature globally, abandoning plans to expand the privacy technology across the platform after years of promoting encrypted communication as the future of messaging.

Apple and Meta are opposing Canada’s proposed Bill C-22, which they say could force companies to weaken encryption or build government-access mechanisms into their products. Canadian authorities argue the bill would help law enforcement respond more quickly to security threats.

After 88 days of nationwide shutdown, internet connectivity has been partially restored in Iran, ending one of the longest full-scale blackouts in modern history. The shutdown was imposed after escalating conflict dynamics and justified by authorities on security grounds. NetBlocks reported that connectivity has largely returned, but that metrics indicate that users still face heavy filtering, including new restrictions on messaging and app stores.

Russian President Vladimir Putin has instructed the government and the FSB to ensure uninterrupted access to essential medical, information, and payment services during periods of restricted mobile internet in Russia. Prime Minister Mikhail Mishustin and FSB Director Alexander Bortnikov are to report back on their progress by 1 July.

Legal

The European Commission has presented a European Technological Sovereignty Package to strengthen Europe’s capacity in semiconductors, AI, cloud infrastructure, and open-source technologies. The package includes two legislative proposals, the Chips Act 2.0 and the Cloud and AI Development Act, alongside an Open Source Strategy and a Strategic Roadmap for Digitalisation and AI in Energy. The proposed Chips Act 2.0, which aims to expand Europe’s semiconductor capabilities and support the next generation of computing infrastructure needed for AI and other advanced technologies. The proposed Cloud and AI Development Act focuses on strengthening Europe’s cloud and AI infrastructure, including data centre capacity, as demand for computing resources rises with the spread of AI. Alongside the legislative proposals, the Commission introduced an Open Source Strategy to support Europe’s software ecosystem and reduce reliance on closed or externally controlled technologies. The Strategic Roadmap for Digitalisation and AI in Energy aims to support the use of digital tools and AI in the energy sector while contributing to a more sustainable digital future.

China has updated its trade secret protection rules to explicitly include digital assets such as data, algorithms, software code, and computer programs. The framework strengthens requirements for corporate safeguards, including confidentiality agreements, employee training, access controls, encryption, and data management practices. It also introduces detailed provisions for digital and remote work environments, encouraging tools like permission-based access, data masking, and activity logging, while explicitly banning illicit methods such as hacking, fraud, and unauthorised system access to obtain trade secrets. The rules further extend protection to overseas infringements that affect China’s domestic market or the rights of Chinese companies.

Meta Platforms is facing growing legal and regulatory pressure both in the USA and Europe over claims that its social media platforms contribute to youth addiction and mental health problems. In New Mexico, the state is seeking $3.7 billion and asking the court to declare Meta a public nuisance. The lawsuit alleges that Facebook, Instagram, and WhatsApp were designed in ways that encourage addictive behaviour among minors. It also claims that these platforms failed to adequately protect young users from harmful content and exploitation. The state is requesting major changes to the platforms, including age verification and restrictions on features such as autoplay and infinite scroll for minors. Meta claims the case concerns individual users, rather than the harm to the public as a whole.

Meta is also attempting to overturn a California jury verdict that found the company negligent in the design of its platforms and awarded damages to a young plaintiff who claimed that social media use contributed to her depression. Meta argues that the claims are barred by Section 230 of the Communications Decency Act and that the alleged harms were connected to online content rather than the platforms’ design features.

In Milan, families and MOIGE have filed a case against Meta and TikTok over alleged failures to protect minors on social media platforms.

The Texas Attorney General has filed a lawsuit against WhatsApp and its parent company, Meta Platforms Inc, alleging that users were allegedly given misleading impressions about the scope and effectiveness of WhatsApp’s end-to-end encryption protections. The lawsuit alleges that Meta and WhatsApp retained broader access to certain user communications and metadata than users may have understood.

Texas has sued Netflix for allegedly collecting user data without consent, including viewing habits, and turning users activity into monetised data. The Attorney General also argues that features such as autoplay were designed to increase engagement, particularly among minors, and that these practices conflict with Netflix’s public representations about being ad-free and family friendly.

The EU Court of Justice upheld an Italian order requiring Meta to compensate publishers for news snippets, intensifying copyright disputes over content reuse and AI training data.

France’s top court rejected Amazon’s challenge to a minimum book delivery fee designed to protect independent bookstores against global platform dominance.

Comisiun na Mean has launched two investigations into Meta under the Digital Services Act, examining whether Facebook and Instagram allow users to effectively choose recommendation feeds that are not based on profiling of their personal data. The regulator also warned that Very Large Online Platforms (VLOPs) must ensure users can exercise their rights under the DSA without manipulation or unnecessary barriers.

A federal jury in California ruled in favour of OpenAI, Sam Altman and Greg Brockman in Elon Musk’s lawsuit alleging that the company abandoned its original non-profit mission after transitioning toward a for-profit structure. The court accepted OpenAI’s argument that Musk had long been aware of discussions around restructuring, concluding that the claims were filed outside the applicable legal time limits and dismissing the case following an advisory jury verdict.

The European Commission issued a fine of €200 million to Temu after finding that Temu’s 2024 risk assessment fell short of the standards laid out in the Digital Services Act (DSA). The Commission further found that Temu did not adequately assess how its platform design, particularly recommender systems and influencer-driven promotion, could contribute to the amplification and spread of illegal products.

Sociocultural

Ofcom has announced a recommendation intended to strengthen protections against illegal intimate image abuse online. The UK regulator said it is updating its Illegal Content Codes to recommend that certain online platforms use automated detection technologies to identify such images, including AI-generated explicit deepfakes and non-consensual image sharing. The recommendation is expected to enter into force in autumn 2026, subject to parliamentary approval.

Similarly, the US Federal Trade Commission has issued guidance for online platforms on compliance with Section 3 of the Take It Down Act, which took effect on 19 May 2026 and requires covered platforms to remove non-consensual intimate photos or videos within 48 hours of receiving a valid request. Covered platforms must provide clear and conspicuous plain-language information about how people can submit removal requests for intimate photos or videos shared without consent. The platforms should make the process easy to use, including for people who do not have an account on the service. The FTC also encourages platforms to help prevent removed images from spreading further, including through hashing technology and, where appropriate, by sharing hashes with services such as the National Center for Missing and Exploited Children’s Take It Down service or StopNCII.org.

Papua New Guinea’s Ministry of Information and Communications Technology has begun drafting instructions and a proposed bill for digital identity and verifiable credentials legislation following the endorsement of the National Digital ID Policy. The proposed legislation will support the national rollout of SevisPass, SevisWallet, SevisDEx, and other approved verifiable credentials. SevisWallet will allow citizens to register, hold, and present trusted digital credentials, while SevisDEx will enable secure, consent-based data exchange.

Dutch citizens and privacy advocates challenged the proposed acquisition of Solvinity, the company hosting the Netherlands’ DigiD identity system, by US-based Kyndryl. Critics warned that allowing an American company near such sensitive infrastructure could expose Dutch citizens’ data to foreign jurisdiction and surveillance risks under US laws. Earlier in May, the District Court of The Hague rejected an attempt by three Dutch citizens to block the government from renewing its contract with Solvinity. However, the Investment Assessment Office (BTI) has advised Willemijn Aerdts, Minister for the Digital Economy and Sovereignty of the Netherlands, to impose a complete ban on this acquisition, as it may pose a risk to the public interest. The minister has acted upon this advice, and the government has blocked the sale.

UNESCO and Pakistan have launched ‘Digital Citizens for Peace’, a media literacy initiative designed to counter hate speech and disinformation by training young journalists and content creators. Through immersive Media and Information Literacy camps, mentorship programmes, and open-access educational toolkits, the initiative aims to strengthen responsible digital engagement and encourage fact-based content creation across the country.

Development

The African Continental Free Trade Area (AfCFTA) Secretariat has selected Kenya, Morocco, and Nigeria as the first countries to implement a digital public infrastructure (DPI) initiative designed to support cross-border trade. The initiative, known as Africa Digital Access and Public Infrastructure for Trade (ADAPT), aims to connect digital identity, trusted data exchange, and interoperable payment systems to reduce friction in intra-African commerce.

Australia has introduced a National Framework for Digital Health Standards to improve interoperability and consistency across healthcare systems. The framework addresses fragmentation caused by independently developed digital health standards. It also provides guidance intended to support coordination between government agencies, healthcare providers, and industry participants. The framework also supports the use of internationally recognised clinical terminology standards and related training initiatives.

Greece’s parliament approved a law which accelerates interoperability between public information systems, establishes a unified digital service platform on gov.gr, introduces a register for entities authorised to use government data-verification services, and strengthens cybersecurity through faster blocking of phishing websites and a national registry of malicious domains.

Economic

The OECD has published a Digital Trade Review of the Association of Southeast Asian Nations, examining regional growth in digital trade and related regulatory challenges.

The OECD said ASEAN benefits from trade openness, increasing digital adoption, and evolving regional policy initiatives. The report noted that uneven participation and fragmented domestic regulations may limit further digital trade integration across the region.

The review identified barriers, including restrictions affecting cross-border data flows, telecommunications, digital services, and trade facilitation systems. The OECD highlighted the importance of regulatory alignment and progress towards paperless trade systems.

The report also discussed opportunities related to AI adoption, including reforms linked to tariffs, data flows, and digital services regulation. These findings underline the importance of coordinated reforms to strengthen ASEAN’s role in the global digital economy.

Pakistan has enacted the Virtual Assets Act 2026, establishing a permanent legal foundation for the Pakistan Virtual Assets Regulatory Authority. The move formalises oversight of the country’s growing digital asset sector, which had previously operated under temporary rules introduced in 2025. Under the new framework, PVARA is responsible for licensing, regulating and supervising virtual assets and virtual asset service providers operating in Pakistan. The law gives the authority powers to issue, suspend and revoke licences, while unlicensed operations can face fines and criminal penalties.

Georgia is preparing to launch GEL₮, a stablecoin linked to the Georgian Lari. According to officials, the initiative is intended to support blockchain-based payment infrastructure under a dedicated stablecoin regulatory framework. Officials said the project could support faster payments, cross-border transfers, fintech development, and programmable financial services.

Brazil’s Central Bank has introduced a new requirement for virtual asset service providers seeking authorisation to operate in the country. From 1 June 2026, companies applying to operate as sociedades prestadoras de serviços de ativos virtuais, or SPSAVs, must submit a reasonable assurance report issued by an independent auditor registered with Brazil’s securities regulator, the Comissão de Valores Mobiliários. The audit requirement is intended to assess whether applicants have adequate compliance and control structures in place. Reviews will focus on anti-money laundering and counter-terrorist financing measures, including governance arrangements, client verification procedures, internal risk controls, and mechanisms to prevent misuse of virtual asset services.

Xi hosts Trump and Putin in two weeks as tech diplomacy takes centre stage

Two high-level summits in as many weeks saw China’s Xi Jinping host first the President of the United States Donald Trump, and then the President of the Russian Federation Vladimir Putin. While the internet has done its thing, comparing Trump’s and Putin’s visits to China right down to the choreography of the children, we’ll focus our analysis on the tech topics that were discussed, with only a brief look at how the two sets of talks compare.

The Trump-Xi Summit signals unease but vague progress

Trade and technology were among the main topics expected to dominate the Trump-Xi summit 13-15 May 2026 in Beijing, the first state visit of a US President to China in 8 years.

A litany of CEOs accompanied Trump on the trip, including Tesla’s Elon Musk, Nvidia’s Jensen Huang, Apple’s Tim Cook, GE Aerospace’s Larry Culp, Boeing’s Kelly Ortberg, Meta’s Dina Powell McCormick, BlackRock’s Larry Fink, Blackstone’s Stephen Schwarzman, Micron’s Sanjay Mehrotra, Mastercard’s Michael Miebach, Qualcomm’s Cristiano Amon and Visa’s Ryan McInerney.

But as much as the public is assured that progress was made, details remain scarce.

When it comes to tech matters, AI was expected to be a central topic of discussion, with reports in the lead-up to the summit suggesting that the USA and China are considering launching formal discussions on AI. ‘The two AI superpowers are going to start talking. We’re going to set up a protocol in terms of how do we go forward with best practices for AI to make sure nonstate actors don’t get a hold of these models,’ US Treasury Secretary Scott Bessent told CNBC.

Part of the AI story are, of course, the chips needed for AI. The USA has tried to limit China’s AI development by restricting sales of advanced chips, with Jensen Huang’s Nvidia and Huang himself caught in the middle. Yet, reports suggest that the USA has cleared around 10 Chinese firms to buy the H200, Nvidia’s second-most powerful AI chip, the H200. However, no deliveries were made. This is likely the breakthrough Huang was seeking in China.

And part of the chips story are the minerals the chips are made of, and on those China has an upper hand which is deftly uses as a lever in its trade relations with the USA. US Trade Representative Jamieson Greer told Bloomberg that rare earth exports from China are improving, although ‌Beijing is still slow to approve some export licences. However, Greer also stated that chip export controls were not a major part of the talks.

A few days later, the White House stated that China will address US concerns regarding supply chain shortages related to rare earths and other critical minerals, as well as concerns over restrictions on the sale of rare earth production and processing equipment and technologies. In response to questions about this statement, China’s Ministry of Commerce said that both sides had discussed the issue and would work to resolve each other’s reasonable and lawful concerns.

Additionally, reports suggest that ByteDance has agreed to purchase millions of Qualcomm ASIC chips for use in its expanding AI infrastructure and agent systems, a deal that might have been struck during the visit.

Russia–China Summit signals deepening cooperation across AI, industry, and digital economy

The Putin-Xi talks on 19-20 May 2026 culminated in a communique that is, off the bat, telling. The document itself is quite long, and while it is not heavy on tech, it still contains important signals.

Putin and Xi emphasised that AI should be developed and deployed in ways that serve universal development rather than narrow national interests. At the same time, they explicitly criticised the use of AI by individual states as a geopolitical instrument to preserve or strengthen global dominance.

The leaders also highlighted growing cooperation on the military dimension of AI. Russia and China intend to deepen collaboration on the development and application of AI in military contexts, both through bilateral channels and within multilateral processes, including discussions under the Group of Governmental Experts on lethal autonomous weapons systems (LAWS).

The two sides committed to coordinating their positions on AI-related scientific and policy questions within international organisations. This indicated an effort to align their approaches more systematically in global standard-setting and norm-shaping forums. It’s not unusual for China and Russia to take similar positions in negotiations over digital matters, but this was a quite explicit geopolitical signal.

China and Russia will merge their GLONASS and Beidou satellite navigation systems into a complementary global service, cooperate in the coordination and use of radio frequencies and satellite orbital slots, and expand cooperation on satellite internet and the internet of things.

Open source cooperation also emerged as a strategic priority. The two countries plan to explore a bilateral software cooperation mechanism and to promote open-source technologies in key industries. That aligns with broader efforts in both countries to localise software ecosystems and reduce reliance on US tech platforms.

The parties agreed to strengthen cooperation in combating cybercrime, welcoming the UN General Assembly’s adoption of the UN Convention against Cybercrime and committing to its rapid entry into force and implementation, including work on an additional protocol to expand its scope and enhance international cooperation.

They pledged to deepen strategic cooperation on information security and coordinated responses to ICT threats. They also intend to exchange experiences in legislative regulation of the internet.

They emphasised the central role of the UN in responding to threats in the information space, after which the countries expressed support for the work of the Global Mechanism. This is also not surprising, considering that the Global Mechanism emerged from the OEWG, which was itself a Russian initiative. The countries also expressed support for the development of broader international legal instruments covering issues such as data security and supply chain resilience. This aligns with China’s longstanding push to establish the Global Initiative on Data Security.

The parties agreed to expand cooperation on the digital economy and cross-border e-commerce, alongside broader collaboration in industries such as automotive, aviation, and mineral extraction. They pledged to strengthen coordination on intellectual property protection, consumer protection in online and cross-border services,

They also agreed to deepen e-commerce cooperation through regional frameworks, including the Eurasian Economic Union (EAEU), the Belt and Road Initiative, the Shanghai Cooperation Organisation (SCO), APEC, and the Enhanced Tumangan Initiative. Cooperation will focus on digitalisation, logistics, barrier-free trade, connectivity, and innovation, with the broader aim of boosting regional economic integration, trade, and employment.

This part is not surprising, considering that China and Russia have repeatedly framed e-commerce, the digital economy, and ICT cooperation as strategic pillars of their partnership since at least the mid-2010s in the SCO and Belt and Road contexts.

Comparing the two summits

There was no joint communique after the Xi–Trump meeting, while the Xi–Putin summit produced a detailed and lengthy one. That alone is quite telling. Xi and Putin also signed over 20 agreements covering energy, trade, science and technology, and infrastructure, while it is still unclear how many agreements Xi and Trump signed.

In the US–China context, comments made by Bessent during Trump’s visit to China, noting that ‘The reason we are able to have wholesome discussions with the Chinese on AI is because we are in the lead,’ highlight continued tech competition between the two countries. Bessent also candidly admitted: ‘I do not think we would be having the same discussions if they were this far ahead of us.’ President Trump also postponed signing the AI and cybersecurity order, saying he did not want new rules to slow US leadership in AI or weaken its competitive edge over China, which reinforces the competitive framing. This is in direct opposition to cooperation with Russia and to future coordination on AI-related scientific and policy questions within international organisations.

China’s engagement with Russia took a more cooperative and forward-looking form. The two sides agreed to actively promote cooperation in joint mineral extraction and in developing green standards. Russia holds some of the world’s largest rare earth reserves—ranked fifth globally—but remains constrained by limited capacity in extraction, refining, and processing, an area where it stands to benefit from Chinese technological and industrial expertise.

Overall, the difference in both outcomes and issue salience indicates that the Russia–China relationship appears more comfortable addressing sensitive domains such as AI cooperation and military applications, whereas US–China engagement remains more constrained and competitive in tone and scope.

Last month in Geneva

Geneva Cyber Week 2026

The UN Institute for Disarmament Research (UNIDIR) and the Swiss Federal Department of Foreign Affairs (FDFA) co-hosted Geneva Cyber Week from 4 to 8 May 2026, bringing policymakers, diplomats, technical experts, industry leaders, academics, and civil society representatives to venues across Geneva and online for a week of discussions on cyber stability, resilience, governance, digitalisation, and the security implications of emerging technologies, including AI.

Returning after its inaugural edition, the event was positioned as a response to a more fragile cyber and geopolitical environment. Held under the theme ‘Advancing Global Cooperation in Cyberspace’, Geneva Cyber Week 2026 took place at a time of mounting cyber insecurity, intensifying geopolitical tensions, and rapid technological change. The programme featured nearly 90 events and reinforced Geneva’s role as a centre for cyber diplomacy, international cooperation, and digital governance.

As part of Geneva Cyber Week, UNIDIR organised the Cyber Stability Conference 2026, on 4–5 May in Geneva and online, bringing together governments, international organisations, industry, academia, and civil society to discuss ICT security and cyber governance. Under the theme ‘Cyber governance in an era of technological revolution: Past lessons, present realities and future frontiers,’ discussions explored how international cyber stability frameworks are adapting to rapid technological change, including AI and quantum computing, while reflecting on lessons from past cyber diplomacy processes and current security challenges.

Multi-year expert meeting on investment, innovation and entrepreneurship for productive capacity-building and sustainable development, 12th session

UNCTAD’s Multi-year Expert Meeting on Investment, Innovation and Entrepreneurship for Productive Capacity-building and Sustainable Development met for its twelfth session on 4-5 May. The experts warned that AI and other strategic technologies are reshaping global investment patterns, concentrating capital in a handful of sectors and countries while leaving many developing economies behind. Discussions at the meeting focused on how developing countries can compete in AI-related sectors, strengthen domestic innovation ecosystems and ensure that AI-driven investment translates into broader development gains.

OSCE Conference ‘Anticipating technologies – for a safe and humane future’ opens in Geneva

The Swiss Chairpersonship of the Organization for Security and Co-operation in Europe opened a two-day high-level conference on anticipatory technologies in Geneva on 7 May. The event examined how foresight, dialogue, and international cooperation can help reduce misunderstandings, build trust, and strengthen security across the OSCE region amid rapid technological change.

The programme included discussions on anticipating technological change and its geopolitical impact, water and energy security in the digital age, and the role of AI in early warning and conflict prevention.

The conference also highlighted Geneva’s role as a meeting point for science and diplomacy, including through institutions such as CERN, the Geneva Science and Diplomacy Anticipator, and the Open Quantum Institute.

The event forms part of the Chairpersonship’s priority to connect scientific and technological anticipation with policy action.

WTO resumed talks as 19 members back e-commerce moratorium pledge

The WTO’s General Council has met in Geneva on 8 May for the first time since the 14th Ministerial Conference (MC14), after negotiators in Yaoundé narrowly missed agreements on several major files, including the future of the long-running e-commerce moratorium and broader WTO reform.

The newly elected chair, Ambassador Clare Kelly, said members remained committed to preserving the careful balance reached during negotiations in Cameroon and avoiding a return to earlier positions.

The discussions follow the expiry on 31 March of the WTO moratorium on customs duties on electronic transmissions, a temporary arrangement first adopted in 1998 that prevented countries from imposing tariffs on digital trade flows such as software downloads, streaming services, and other online transmissions. Ministers at MC14 failed to agree on another extension, exposing deep divisions over how long the moratorium should continue and how governments should respond to rapid technological developments such as AI and 3D printing.

Despite the lapse, a group of 19 WTO members announced they would continue not imposing customs duties on electronic transmissions among themselves. In a joint statement circulated at the WTO, the group said the arrangement would provide predictability and certainty for businesses and consumers while multilateral negotiations continue. The group includes Argentina, Australia, Costa Rica, Ecuador, Guatemala, Iceland, Israel, Japan, South Korea, Malaysia, Mexico, New Zealand, Norway, Panama, Paraguay, Singapore, Separate Customs Territory of Taiwan, Penghu, Kinmen and Matsu, the USA, and Uruguay.

Türkiye also signalled new flexibility during the General Council meeting, announcing it would not block consensus on a temporary extension of the moratorium. However, Brazil maintained its opposition to a four-year extension of the moratorium.

The chair announced further consultations on e-commerce and WTO reform, with plans to report back to members in July.

Intergovernmental Group of Experts on E-commerce and the Digital Economy 9th session

The UNCTAD Intergovernmental Group of Experts on E-commerce and the Digital Economy met in Geneva for its 9th session from 11–13 May 2026 to address how digital trade affects fiscal systems in developing countries. The agenda focused on strengthening tax revenues in the context of e-commerce, including VAT and customs challenges, platform-based taxation, and the implications of cross-border digital transactions. Sessions explored revenue risks, administrative digital tools, and the role of platforms as tax intermediaries, alongside international and regional cooperation to improve tax readiness. The meeting also covered measuring the digital economy and advancing capacity-building to support inclusive and sustainable digital development strategies.

UN Virtual Worlds Day 2026

The 3rd UN Virtual Worlds Day, co-organised by the International Telecommunication Union (ITU) and multiple UN agencies and partners, took place in Geneva on 11–12 May 2026. It convened ambassadors, ministers, city leaders, UN entities, industry pioneers and innovators for two days of high-level dialogue on AI-driven cities, immersive environments and spatial intelligence. Discussions focused on how frontier technologies are reshaping urban governance and public services, including the emerging citiverse concept and its link to the Global Digital Compact. It concluded with an executive briefing on AI, spatial intelligence and AI-enabled citiverse, and the Call to Action for inclusive digital futures.

79th World Health Assembly

The 79th World Health Assembly (WHA79), which took place in Geneva from 18–23 May 2026, brought together WHO member states, international organisations, civil society, and health stakeholders to discuss global health priorities and governance. Alongside the formal Assembly proceedings, a wide range of side events focused on digital health and AI, covering topics such as trusted health data, interoperability, responsible AI governance, and digital transformation of health systems. Discussions examined how AI and digital technologies can support universal health coverage, strengthen evidence-based policymaking, improve diagnostics and ageing care, and enhance cross-border cooperation, while also addressing issues of ethics, transparency, data governance, and equity in healthcare systems.

UN human rights experts raise concerns over the large-scale sexual exploitation of women and girls online

The UN Special Rapporteur on violence against women and girls, its causes and consequences and the UN Special Rapporteur on the right to privacy issued a statement raising concerns over what they describe as large-scale sexual exploitation of women and girls facilitated and monetised through Pornhub and its parent company, Aylo Holdings, as well as by payment networks and search engines that enable distribution and monetisation of content.

The rapporteurs focused on Pornhub and its parent company Aylo Holdings, while also referencing broader concerns involving user-generated pornography platforms, payment networks, and search and technology companies linked to online distribution systems.

The experts said businesses involved in digital content ecosystems should not avoid responsibility where their services contribute to human rights violations. They called for stronger safeguards, including mandatory third-party age and consent verification systems for user-generated pornography platforms.

UN Human Rights Office issues guidelines on child safety online

A new brief by the Office of the UN High Commissioner for Human Rights (OHCHR) outlined 10 interconnected steps to improve children’s online safety through a rights-based, system-level regulatory approach. Rather than relying on broad restrictions or shifting responsibility to users, it prioritises structural changes to how platforms are designed and governed.

First, it calls for safety-by-design obligations that reduce harmful features such as autoplay, infinite scroll, and intrusive notifications. Regulation should be grounded in children’s rights, ensuring that protection is balanced with rights to participation, expression, privacy, and access to information. Strong data privacy protections are central, including prohibiting micro-targeting and limiting the profiling of children for commercial purposes.

Companies are expected to carry out ongoing child rights impact assessments as part of broader human rights due diligence, with transparency and accountability for their findings. Where age verification is used, it must include strict safeguards around data collection, storage, and non-discrimination.

The guidance emphasises targeted, harm-specific age restrictions rather than blanket bans, and stresses that children should be actively involved in shaping the policies that affect their digital environments. Platforms should also be required to provide meaningful transparency on recommender systems, moderation practices, and design choices.

Finally, effective oversight, enforceable accountability mechanisms, and continuous, evidence-based policymaking are essential to ensure that regulation adapts as technologies and risks evolve.

Notably, the guidance frames online child safety as both a governance issue and a rights issue. It argues that children’s rights apply online as fully as they do offline, including rights related to participation, privacy, access to information, and freedom of expression. Children, the brief notes, should also have a say in how digital spaces are governed and regulated.

Don’t miss!

Follow the road to the 2027 Geneva AI Summit, a global journey shaping AI’s future. Bookmark the page and register for updates to stay informed and involved every step of the way to Geneva 2027.

HIGHLIGHT OF THE WEEK

IN OTHER NEWS LAST WEEK

LOOKING AHEAD

LAST WEEK IN GENEVA

READING CORNER

DON’T MISS

HIGHLIGHT OF THE WEEK

Diplo at Global Dialogue on AI governance, WSIS and AI for Good

IN OTHER NEWS LAST WEEK

LOOKING AHEAD

READING CORNER

DON’T MISS

HIGHLIGHT OF THE WEEK

IN OTHER NEWS LAST WEEK

LAST WEEK IN GENEVA

LOOKING AHEAD

READING CORNER

DON’T MISS

June 2026 in retrospect

AI governance in June

—

Geneva at the centre of AI governance: Where technology, diplomacy, and humanity converge

A dense ecosystem of global governance

Between technology, diplomacy, and intellectual tradition

Geneva as the centre for AI diplomacy

Beyond events: An emerging governance architecture

Don’t miss!

The Fable 5 case and the problem of reactive AI governance

Beyond AI: The developments that made waves in June

Beyond AI: The developments that made waves in June

—

Africa’s digital diplomacy: Building a common voice for global digital governance

‘Tech Liberation Day’: EU launches tech sovereignty push

HIGHLIGHT OF THE WEEK

IN OTHER NEWS LAST WEEK

LAST WEEK IN GENEVA

READING CORNER

DON’T MISS

HIGHLIGHT OF THE WEEK

IN OTHER NEWS LAST WEEK

LAST WEEK IN GENEVA

LOOKING AHEAD

READING CORNER

DON’T MISS

HIGHLIGHT OF THE WEEK

IN OTHER NEWS LAST WEEK

LAST WEEK IN GENEVA

LOOKING AHEAD

READING CORNER

DON’T MISS

Le mois dernier en matière de gouvernance de l’IA

Gouvernance mondiale de l’IA

Partenariats

Stratégies nationales en matière d’IA

Cybersécurité

Droits de l’Homme

Juridique

Économie

Socioculturel

Gouvernance des contenus

Renforcement des capacités

Magnifica humanitas : la dignité humaine à l’ère de l’IA

Au-delà de l’IA : les développements qui ont fait sensation en mai

Technologies

Cybersécurité

Droits de l’Homme

Juridique

Socioculturel

Développement

Économie

Xi reçoit Trump et Poutine en l’espace de deux semaines alors que la diplomatie technologique occupe le devant de la scène

Le sommet Trump-Xi témoigne d’un certain malaise, mais laisse entrevoir de vagues progrès

Le sommet Russie-Chine annonce un renforcement de la coopération dans les domaines de l’IA, de l’industrie et de l’économie numérique

Comparaison des deux sommets

Le mois dernier à Genève

Geneva Cyber Week 2026

Réunion d’experts pluriannuelle sur l’investissement, l’innovation et l’entrepreneuriat pour le renforcement des capacités productives et le développement durable, 12e session

Ouverture à Genève de la conférence de l’OSCE intitulée « Anticiper les technologies – pour un avenir sûr et humain »

L’OMC a repris les négociations alors que 19 membres soutiennent l’engagement en faveur d’un moratoire sur le commerce électronique