The company said the controls were applied on 12 June to Claude Fable 5 and Claude Mythos 5, requiring access restrictions for foreign nationals inside and outside the United States. Anthropic suspended access to both models for all users because it said it had no reliable way to verify nationality in real time.
Anthropic said the controls were lifted on 30 June. Fable 5 will become available globally from 1 July on the Claude Platform, Claude.ai, Claude Code and Claude Cowork, with access on AWS, Google Cloud and Microsoft Foundry to be restored as quickly as possible.
Access to Mythos 5 has been restored only for a set of US organisations following government approval. Anthropic said Fable 5 and Mythos 5 share the same underlying model, but Fable 5 has stronger safeguards for general use, while Mythos 5 has fewer safeguards and is limited to trusted partners working on defensive cybersecurity.
The export control directive followed a report by Amazon researchers describing a method for bypassing Fable 5 safeguards. Anthropic said the reported behaviour involved identifying software vulnerabilities and, in one case, producing code showing how a vulnerability could be exploited.
The company said its review found that the technique did not expose unique Mythos-level cyber capabilities. It has trained an improved safety classifier to block the behaviour described in the report, and said blocked requests will be redirected to Claude Opus 4.8.
Anthropic also called for a shared industry framework to assess the severity of AI jailbreaks. It said it is working with Amazon, Microsoft, Google and other Glasswing partners on criteria including capability gain, breadth of capability gain, ease of weaponisation and discoverability.
The company said it is expanding cooperation with the US government on frontier AI security, including pre-release evaluation, faster information sharing and joint research on safeguards.
Why does it matter?
The case shows how frontier AI releases are becoming part of national security and export-control policy, especially when models have advanced cybersecurity capabilities. Anthropic’s response also highlights a broader governance gap: governments and companies still lack a shared standard for judging when a jailbreak is minor, serious or urgent enough to justify intervention. The outcome could influence how advanced AI models are tested, released and restricted across borders.
Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!
The alliance said age restrictions on mainstream social media platforms could reduce some risks. Still, children may move to less regulated digital spaces, including encrypted messaging services, gaming platforms and other online environments where grooming, sexual extortion and abuse can continue.
UK ATOC called for a broader, system-wide response focused on prevention, stronger platform accountability and safer-by-design digital services. It said governments, regulators, technology companies and online service providers share responsibility for reducing opportunities for abuse before harm occurs.
The alliance proposed a package of technical, legislative and regulatory measures. These include stronger safeguards in end-to-end encrypted environments, robust age-assurance systems, mandatory safer-by-design principles, stronger enforcement under the Online Safety Act and clearer regulation of AI chatbots and companion services.
It also called for device-level nudity detection, upload prevention for known child sexual abuse material and measures to address livestreamed abuse, grooming and sexual extortion.
UK ATOC welcomed the government’s plan to introduce nudity-detection tools on children’s devices, describing it as an important additional safeguard.
The UK debate shows the limits of age-based social media bans as a child-safety tool. Online child sexual exploitation and abuse can move across platforms, devices, encrypted services, gaming environments and AI-enabled systems. UK ATOC’s response therefore shifts the focus from access restrictions alone towards prevention, safer design, platform duties and technical safeguards that address how abuse actually happens across digital services.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
The OECD has published a report examining how AI could support citizen participation and democratic innovation while highlighting the safeguards needed for its responsible use.
The report, Artificial Intelligence and the Future of Citizen Participation, was approved and declassified by the OECD Public Governance Committee on 22 June 2026. It was produced as part of the OECD Public Governance Reviews series in collaboration with the Bertelsmann Stiftung.
The report says public participation can help governments design better policies and strengthen trust. It cites OECD trust findings showing that people who feel they have a say in government decisions are far more likely to report high trust in government.
The OECD notes that governments have long relied on digital technologies, including online platforms and civic tech tools, to expand public participation. AI represents the next stage of this evolution, with governments increasingly experimenting with tools for consultation, deliberation, communication and policy analysis.
The report is based on desk research and analysis of 50 AI use cases in participation processes from 22 OECD member and partner countries. It proposes a typology to help public officials and practitioners understand where AI tools may be useful and what challenges they may address.
Based on an analysis of 50 AI use cases from 22 OECD member and partner countries, the report proposes a typology covering nine categories of AI applications, including information development, sense-making, translation, transcription, virtual assistance, moderation, facilitation, simulation and participation architecture.
These tools can support both front-office activities, where citizens interact directly with government, and back-office activities, where public administrations design, analyse and manage processes internally.
According to the OECD, AI could make participation processes more accessible and efficient by helping governments analyse large volumes of public input, improve communication, reduce administrative costs and broaden participation.
Sense-making tools can help analyse large amounts of text submitted during consultations. Translation and transcription tools can make processes more accessible across languages and formats, while virtual assistants can help people navigate information about citizen participation opportunities.
AI can also support moderation and facilitation. The report says such tools may help prevent spam, hate speech or manipulation in online discussions, and could support live deliberation by identifying common ground or structuring debate.
However, the OECD cautions against treating AI as a simple fix for democratic challenges. It says technology alone cannot solve problems such as weak links between participation processes and actual policy decisions.
The report also highlights ethical, operational and societal risks, including algorithmic bias, opaque decision-making, hallucinations, cybersecurity threats, digital exclusion and declining public trust if AI systems are poorly designed or deployed.
The OECD also highlights the risks of inaction, noting that governments may miss valuable opportunities if they avoid AI tools even when they could be applied responsibly.
The report says governments should establish guardrails for AI use in citizen participation, including transparency, compliance with democratic values, protection of civic space, attention to data divides and low-tech alternatives for citizens with limited digital access.
It also calls for stronger enablers, including AI literacy, skills development, citizen engagement in the design and governance of AI systems, open standards where appropriate, and support for scaling successful pilots.
The OECD concludes that most public-sector use of AI in citizen participation remains experimental. It argues that lasting benefits will depend on transparent governance, human oversight and continued efforts to strengthen democratic participation beyond technology alone.
Why does it matter?
Governments are increasingly exploring AI as a way to make public participation more accessible, scalable and responsive. The OECD’s report shows that AI can support consultation, deliberation and policy analysis, but only when accompanied by safeguards that protect transparency, inclusion and democratic accountability.
The report also reinforces a broader shift in AI governance from technical capability to institutional design. By emphasising human oversight, civic participation, digital inclusion and democratic values, the OECD argues that AI should enhance, not replace, the processes that underpin public trust and democratic decision-making.
Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!
The Organisation for Economic Co-operation and Development (OECD) has published a policy paper examining how AI and digital tools can help small and medium-sized enterprises (SMEs) gain better access to sustainable finance, where they remain significantly underrepresented.
The paper maps practical applications of AI and digital tools across the entire financing lifecycle, from sustainability data generation and reporting by SMEs to loan origination, credit assessment and portfolio monitoring by financial institutions. The OECD notes that AI has the potential to support the front, middle and back office of lending operations rather than a single stage of the financing process.
Drawing on country examples and recent initiatives, the OECD argues that technological adoption must be accompanied by appropriate governance. It identifies four policy priorities: developing interoperable data infrastructure, strengthening verification mechanisms, creating incentives for SME sustainability reporting and ensuring accountable use of AI in financing decisions.
Why does it matter?
Small and medium-sized enterprises account for much of economic activity and employment but often struggle to access sustainable finance because they lack the resources to produce the data and reporting required by lenders and investors. AI could reduce these costs by automating data collection, reporting and credit assessment, making green finance more accessible to smaller businesses.
The OECD also emphasises that technology alone will not close the financing gap. Real progress depends on reliable data infrastructure, effective verification and clear governance to ensure AI-supported financing decisions are transparent, accountable and fair, preventing existing inequalities from being reinforced.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
Africa’s place in an evolving digital governance landscape
As AI, cybersecurity, and digital technologies become increasingly central to international policymaking, African countries are seeking to strengthen their role in shaping global digital governance. Questions of representation, digital sovereignty, capacity development, and regional coordination are becoming more prominent as governments prepare for negotiations on AI governance, cybersecurity, telecommunications, and internet governance.
These issues formed the focus of a recent Diplo webinar on Cyber Diplomacy in Africa: Regional, National and Continental Initiatives, moderated by Mwende Njiraini, African Initiative Coordinator at Diplo and Chair of the ITU-T Study Group 17 Regional Group for Africa on security. The discussion brought together policymakers, diplomats, academics, and regional organisations to examine how African interests can be more effectively represented in international digital governance processes.
Although the discussion focused on Africa, many of the issues raised, including AI governance, digital sovereignty, capacity development, and multistakeholder cooperation, reflect broader challenges facing digital governance worldwide.
From cyber diplomacy to diplomacy in the AI era
Opening the discussion, Kurbalija suggested that the distinction between cyber diplomacy, digital diplomacy, and technology diplomacy is becoming less significant as digital technologies permeate virtually every area of international relations. Rather than focusing on terminology, he argued that the central question is how countries, communities, and citizens represent their interests in an increasingly digital world.
‘Cyber diplomacy, digital diplomacy, or AI diplomacy is ultimately diplomacy. It is about representing interests, negotiating, and finding common solutions.’, he said.
According to Kurbalija, technological developments are no longer confined to specialised policy discussions. AI, cybersecurity, digital infrastructure, and data governance increasingly influence trade, security, education, healthcare, humanitarian action, and economic development, making digital issues part of mainstream diplomacy.
This evolution also raises questions about whether Africa is sufficiently represented in international discussions shaping the future of digital technologies.
Image via Magnific
Kurbalija noted that African diplomats are becoming more active in negotiations related to AI, cybersecurity, and internet governance, but argued that stronger participation will be necessary to ensure that the continent’s priorities are reflected in emerging international frameworks.
Rather than approaching these meetings individually, Kurbalija encouraged participants to prepare coordinated positions that reflect African priorities across different policy areas.
Regional coordination remains a work in progress
A recurring theme throughout the discussion was the gap between continental ambitions and national implementation.
However, she questioned whether these processes consistently translate into practical outcomes across the continent.
To illustrate this point, Getao presented the results of a live audience poll measuring familiarity with African digital governance initiatives. While approximately half of the participants recognised the AU Convention on Cyber Security and Personal Data Protection (the Malabo Convention), significantly fewer were familiar with other continental initiatives, including the AU Digital Transformation Strategy and the African Union’s position on international law in cyberspace.
The findings suggested that awareness of Africa’s existing digital governance architecture remains uneven, even among participants engaged in digital policy discussions.
Ambassador Bitange Ndemo argued that implementation presents an even greater challenge than awareness. He observed that agreements adopted at the African Union level often take considerable time to influence national policymaking, with countries frequently developing their own legal and regulatory approaches rather than building on common continental frameworks.
Using the Malabo Convention as an example, Ndemo suggested that many governments introduced separate data protection legislation without fully integrating broader continental approaches. According to him, one contributing factor is reliance on external funding for many regional digital initiatives.
‘If we continue depending on external partners to finance our priorities, ownership becomes more difficult’, Ndemo added.
Ndemo argued that stronger African investment in digital governance initiatives would improve both implementation and long-term sustainability.
Getao echoed this concern, noting that important achievements at the continental level do not always ‘percolate’ effectively to national implementation.
Building common African positions
Despite these challenges, speakers highlighted several examples of growing regional coordination.
Meriem Slimani described how the African Telecommunications Union (ATU) has worked to strengthen cooperation among member states in preparing common African positions for international telecommunications negotiations.
When she joined the organisation in 2015, Slimani recalled, many countries submitted proposals independently at international meetings, often without consulting neighbouring states.
ATU responded by creating a coordination platform through which member countries discuss priorities, identify common interests, exchange experiences, and gradually develop shared positions before major international conferences.
‘Our objective has been to ensure that Africa speaks with one voice where common interests exist.’
Image via Magnific
According to Slimani, this collaborative approach has become particularly important in preparation for major meetings of the International Telecommunication Union (ITU), where coordinated regional positions can strengthen Africa’s influence during negotiations.
While acknowledging that implementation challenges remain, he argued that progress has been more visible in some sectors than others.
In particular, COMESA has advanced several practical digital trade initiatives, including electronic trade documentation, digital logistics systems, electronic certificates of origin, and simplified digital trade procedures designed to facilitate cross-border commerce.
Governance issues such as cybersecurity and cybercrime, however, have generally progressed more slowly because they often involve more politically sensitive discussions and require broader legal coordination among participating states.
Chinemhute suggested that smaller regional organisations can sometimes move more quickly than continental institutions because they involve fewer actors and more focused policy priorities.
Looking ahead
While speakers approached Africa’s digital future from different institutional and regional perspectives, several common priorities emerged throughout the discussion. These included strengthening Africa’s participation in global digital governance processes, improving coordination among national, regional, and continental initiatives, investing in capacity development, and ensuring that digital policies reflect local realities and priorities.
The discussion also highlighted that digital governance extends beyond technology. Questions of AI, cybersecurity, connectivity, language, education, and financing were presented as interconnected challenges that require cooperation among governments, regional organisations, academia, the private sector, and civil society.
Image via Magnific
As international discussions on AI and digital governance continue through forums such as the AI for Good Global Summit, the World Summit on the Information Society (WSIS)+20 process, and the Internet Governance Forum (IGF), speakers stressed that African participation will be most effective when supported by coordinated regional positions and sustained investment in local expertise and digital capabilities.
Ultimately, the webinar underscored that Africa’s role in shaping the future of digital governance will depend not only on engagement in international negotiations but also on translating continental ambitions into practical national implementation and ensuring that African perspectives contribute to global debates on AI, cybersecurity, and digital development.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
The European Commission had proposed replacing cookie banners with an automated browser-based privacy signal as part of its ‘Digital Omnibus’ package, a move that would have allowed devices to communicate users’ tracking preferences directly to websites. The plan, outlined in Article 88b of the GDPR, was intended to cut red tape and reduce the burden on consumers navigating consent requests across the web.
According to digital rights organisation noyb, cookie banners were not created by data protection law but emerged as a mechanism for the online advertising industry to obtain users’ consent for data sharing with third parties. Studies suggest only 3 to 10 per cent of users actually wish to be tracked, yet so-called dark patterns, such as hidden ‘no’ buttons and pre-ticked boxes, allow the industry to achieve consent rates of up to 90 per cent. Across more than 450 million EU citizens, this results in billions of unnecessary clicks each year.
According to noyb, a lobbying document submitted by Google argued that removing cookie banners would effectively halt all online advertising, citing figures that the European Commission has since described as highly exaggerated. The Commission had made clear that consent would still be possible on a per-website and per-purpose basis, meaning users could grant access to specific outlets while withholding it from others. Google’s paper also claimed that media outlets would be harmed, despite the fact that they are explicitly exempt from the proposed provision.
According to noyb, the lobbying campaign appears to have influenced the legislative process. In the Council’s position paper of 18 June 2026, Article 88b was removed entirely from the Digital Omnibus. Noyb added that Germany, France, and Poland were among the member states supporting the article’s removal following lobbying by the online advertising industry.
The outcome is particularly striking given that many of the same member states have long called on the EU to simplify regulation and cut red tape. noyb, the European digital rights organisation, has described the result as a victory for lobbying over public interest, noting that the majority of EU citizens have consistently expressed frustration with cookie banners.
The European Parliament has not yet taken a position on Article 88b, and negotiations between the Parliament and the Council are ongoing. Noyb has urged the European Parliament to support reinstating Article 88b during the next stage of negotiations.
Why does it matter?
The debate highlights the growing tension between digital simplification efforts, privacy protection and the economic interests of the online advertising ecosystem. Browser-based privacy signals have long been discussed as a way to reduce repetitive consent requests while preserving users’ ability to decide when and how their personal data may be used.
The proposal’s removal also illustrates the influence that industry stakeholders can have during the EU legislative process. Whether Article 88b is reinstated during negotiations with the European Parliament could shape the future of online consent management in Europe, affecting digital advertising, user experience and the practical implementation of data protection rules.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
Greece has opened a public consultation on a draft law to implement the EU AI Act and create a national framework for AI governance.
The Ministry of Digital Governance and Artificial Intelligence said the draft law has been under public consultation since 21 June 2026, with comments open until 6 July. The proposal aims to introduce the national mechanisms needed to apply the AI Act in Greece while supporting innovation, competitiveness and the protection of fundamental rights.
Under the draft law, the Hellenic Data Protection Authority would become the central market surveillance authority and national contact point for AI Act implementation. The Hellenic Telecommunications and Post Commission would act as the notifying authority for conformity assessment procedures.
The proposal would also establish an Artificial Intelligence Coordination and Expertise Centre to support the implementation of the new framework.
It would create an AI regulatory sandbox, allowing startups and small and medium-sized enterprises to develop and test innovative AI applications in real-world conditions with support from the state.
The draft law also introduces a complaint-handling mechanism, an administrative sanctions system and a unified registry of AI systems used by public-sector bodies. The registry is intended to strengthen transparency, accountability and public trust in government use of AI.
Greece’s proposal shows how the EU AI Act is moving from Brussels-level legislation into national enforcement structures. The draft law would assign supervisory roles, create a national AI coordination centre, establish a regulatory sandbox and require a public-sector AI registry. Such measures could shape how AI systems are monitored, tested and deployed across both government and the private sector, while giving startups clearer pathways for compliance.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
The UK’s Information Commissioner’s Office (ICO) has warned that unauthorised access to patient records is a serious breach of trust and an ongoing concern across the healthcare sector. In a new blog, the regulator said medical records contain some of the most sensitive personal information and must only be accessed for legitimate reasons.
The ICO said inappropriate access remains rare and does not reflect the behaviour of most healthcare professionals. However, recent high-profile incidents suggest the problem is not confined to isolated cases and requires a stronger organisational response.
According to the regulator, personal curiosity is never a legitimate basis for accessing patient records. Deliberate or reckless access to personal data without authorisation is unlawful and may result in disciplinary measures, loss of professional registration and, in some cases, criminal prosecution.
The ICO called on healthcare leaders to strengthen organisational culture through clear communication, role-specific data protection training and technical safeguards, including role-based access controls and audit logging. Protecting patient privacy is fundamental to maintaining trust in the healthcare system in the UK.
Why does it matter?
Healthcare records contain some of the most sensitive categories of personal information, including medical histories, diagnoses and treatment details. Even isolated cases of unauthorised access can undermine public trust in healthcare institutions and raise concerns about privacy, confidentiality and professional accountability.
The warning also highlights the growing importance of data governance in healthcare. As health systems become increasingly digital and interconnected, organisations must combine technical safeguards, staff training and strong organisational culture to ensure sensitive information is accessed only when necessary and for legitimate purposes. Maintaining patient trust remains essential to the effective delivery of healthcare services.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
Ofcom has imposed an £80,000 fine on pornography provider First Time Videos LLC after finding that the company failed to implement legally required age assurance measures under the Online Safety Act.
Alongside the enforcement action, Ofcom announced its provisional view that xgroovy.com may also have failed to comply with age assurance obligations under the legislation. The regulator further expanded an existing investigation into Sun Social Media Inc. to cover an additional adult website operated by the company.
Ofcom said the penalty was determined with regard to the size and turnover of the service, ensuring that the sanction remained proportionate while reinforcing compliance expectations across the sector.
Why does it matter?
The decision marks an important milestone in the implementation of the UK’s Online Safety Act, demonstrating that age assurance requirements are moving beyond policy commitments into active regulatory enforcement. By imposing financial penalties on non-compliant providers, Ofcom is signalling that online platforms hosting adult content will be expected to adopt effective measures to prevent children’s access.
The case also reflects a broader international trend towards stronger child online safety regulation. Governments and regulators increasingly view age assurance technologies as a key tool for protecting minors in digital environments, while balancing concerns around privacy, proportionality and implementation. Future enforcement actions could shape how platforms design and deploy age verification systems both in the UK and beyond.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
The European Parliament has approved amendments to parts of the EU AI Act as part of the digital omnibus package, postponing some compliance deadlines while adding a ban on AI systems used to create non-consensual sexually explicit content.
MEPs backed the changes with 423 votes in favour, 57 against and 174 abstentions. The measures are intended to simplify compliance for companies while preserving the AI Act’s risk-based structure and core safeguards.
Under the approved text, obligations for stand-alone high-risk AI systems would apply from 2 December 2027. Obligations for AI systems embedded as safety components in products covered by the EU sectoral safety and market surveillance legislation would apply from 2 August 2028.
The text also delays the obligation to watermark AI-generated content until 2 December 2026. By then, AI-generated content will need to be labelled in a machine-readable way to support transparency.
Parliament also approved a ban on AI systems that generate child sexual abuse material or create images, videos or audio depicting an identifiable person’s intimate parts or sexually explicit activities without consent. Providers would not be allowed to place such systems on the EU market unless they include adequate technical safeguards to prevent the creation of such material. The ban would also apply to deployers using systems for that purpose.
Other changes include removing overlapping requirements for AI used in machinery products, clarifying the definition of ‘safety component’, extending some SME exemptions to small mid-cap enterprises, and streamlining enforcement of certain general-purpose AI systems through the EU AI Office.
The legislation still needs formal adoption by the Council before it can enter into force.
Why does it matter?
The vote shows the EU trying to adjust the implementation AI the AI Act without reopening the law’s overall risk-based architecture. Delaying some deadlines could reduce legal uncertainty for businesses and give standards, guidance, and support measures more time to mature. At the same time, the proposed ban on nudification tools and AI-assisted child sexual abuse material addresses a fast-growing harm linked to generative AI, especially image and video manipulation targeting women and minors.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
