Competing visions of AGI emerge at Google DeepMind and Microsoft

Two former DeepMind co-founders now leading rival AI labs have outlined sharply different visions for how artificial general intelligence (AGI) should be developed, highlighting a growing strategic divide at the top of the industry.

Google DeepMind chief executive Demis Hassabis has framed AGI as a scientific tool for tackling foundational challenges. These include fusion energy, advanced materials, and fundamental physics. He says current models still lack consistent reasoning across tasks.

Hassabis has pointed to weaknesses, such as so-called ‘jagged intelligence’. Systems can perform well on complex benchmarks but fail simple tasks. DeepMind is investing in physics-based evaluations and AlphaZero-inspired research to enable genuine knowledge discovery rather than data replication.

Microsoft AI chief executive Mustafa Suleyman has taken a more product-led stance, framing AGI as an economic force rather than a scientific milestone. He has rejected the idea of race, instead prioritising controllable and reliable AI agents that operate under human oversight.

Suleyman has argued that governance, not raw capability, is the central challenge. He has emphasised containment, liability frameworks, and certified agents, reflecting wider tensions between rapid deployment and long-term scientific ambition as AI systems grow more influential.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

Russia considers restoring Roblox access after compliance talks

Roblox has signalled willingness to comply with Russian law, opening the possibility of the platform being unblocked in Russia following earlier access restrictions.

Roskomnadzor stated that cooperation could resume if Roblox demonstrates concrete steps instead of declarations towards meeting domestic legal requirements.

The regulator said Roblox acknowledged shortcomings in moderating game content and ensuring the safety of user chats, particularly involving minors.

Russian authorities stressed that compliance would require systematic measures to remove harmful material and prevent criminal communication rather than partial adjustments.

Access to Roblox was restricted in early December after officials cited the spread of content linked to extremist and terrorist activity.

Roskomnadzor indicated that continued engagement and demonstrable compliance could allow the platform to restore operations under the regulatory oversight of Russia.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

OpenAI expands AI training for newsrooms worldwide

The US tech company, OpenAI, has launched the OpenAI Academy for News Organisations, a new learning hub designed to support journalists, editors and publishers adopting AI in their work.

An initiative that builds on existing partnerships with the American Journalism Project and The Lenfest Institute for Journalism, reflecting a broader effort to strengthen journalism as a pillar of democratic life.

The Academy goes live with practical training, newsroom-focused playbooks and real-world examples aimed at helping news teams save time and focus on high-impact reporting.

Areas of focus include investigative research, multilingual reporting, data analysis, production efficiency and operational workflows that sustain news organisations over time.

Responsible use sits at the centre of the programme. Guidance on governance, internal policies and ethical deployment is intended to address concerns around trust, accuracy and newsroom culture, recognising that AI adoption raises structural questions rather than purely technical ones.

OpenAI plans to expand the Academy in the year ahead with additional courses, case studies and live programming.

Through collaboration with publishers, industry bodies and journalism networks worldwide, the Academy is positioned as a shared learning space that supports editorial independence while adapting journalism to an AI-shaped media environment.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Google launches Gemini 3 Flash for scalable frontier AI

The US tech giant, Google, has unveiled Gemini 3 Flash, a new frontier AI model designed for developers who need high reasoning performance combined with speed and low cost.

Built on the multimodal and agentic foundations of Gemini 3 Pro, Gemini 3 Flash delivers faster responses at less than a quarter of the price, while surpassing Gemini 2.5 Pro across several major benchmarks.

The model is rolling out through the Gemini API, Google AI Studio, Vertex AI, Android Studio and other developer platforms, offering higher rate limits, batch processing and context caching that significantly reduce operational costs.

Gemini 3 Flash achieves frontier-level results on advanced reasoning benchmarks while remaining optimised for large-scale production workloads, reinforcing Google’s focus on efficiency alongside intelligence.

Early adopters are already deploying Gemini 3 Flash across coding, gaming, deepfake detection and legal document analysis, benefiting from improved agentic capabilities and near real-time multimodal reasoning.

By lowering cost barriers while expanding performance, Gemini 3 Flash enhances Google’s competitive position in the rapidly evolving AI model market. It broadens access to advanced AI systems for developers and enterprises.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

OpenAI brings in former UK chancellor George Osborne

Former UK chancellor George Osborne has joined OpenAI in a London-based role. He will lead the OpenAI for Countries programme focused on government partnerships.

The initiative aims to help governments build AI capacity and ensure systems reflect democratic values. OpenAI says more than 50 countries are already involved.

Osborne will work on developing AI infrastructure, boosting AI literacy and improving public services. The role follows discussions with OpenAI chief executive Sam Altman.

His appointment comes as UK-US tech talks face setbacks and investment in AI accelerates. Against this backdrop, financial authorities have warned of risks linked to the sector’s rapid growth.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

AI-generated ads face new disclosure rules in South Korea

South Korea will require advertisers to label AI-generated or AI-assisted advertising from early 2026, marking a shift in how the country governs AI in online commerce and consumer protection.

The measure responds to a sharp rise in deceptive ads using synthetic imagery and deepfakes, particularly in healthcare and financial promotions. Regulators say transparency at the point of content delivery is intended to reduce manipulation and restore consumer trust.

Authorities in South Korea acknowledge that mandatory labelling alone may not deter malicious actors, who can bypass rules through offshore hosting or rapidly changing content. Detection challenges and uneven enforcement capacity across platforms remain open concerns.

South Korea’s industry groups warn that the policy could have uneven economic effects within the country’s advertising ecosystem. Large platforms and agencies are expected to adapt quickly, while smaller firms may face higher compliance costs that slow experimentation with generative tools.

Policymakers argue the framework aligns with South Korea’s broader AI governance strategy, positioning the country between innovation-led and precautionary regulatory models as synthetic media becomes more widespread.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

Segment Anything adds audio as Meta unveils SAM Audio

Meta has introduced SAM Audio, a new AI model that uses intuitive prompts to isolate and segment sounds from complex audio recordings. The release extends the company’s Segment Anything collection beyond visuals into audio and video workflows.

SAM Audio allows users to separate sounds through text prompts, visual cues, or time-based selections. Creators can extract vocals or instruments, remove background noise, or isolate specific sound sources in recordings without specialised audio engineering tools.

Meta describes SAM Audio as a unified model designed around how people naturally think about sound. It supports combined text, visual, and time-based prompts, enabling flexible audio separation across music, podcasting, film, accessibility, and research.

Meta says the model achieves strong performance across diverse audio environments and is already being used internally to develop next-generation creative tools. The approach lowers technical barriers while expanding the range of possible audio editing applications.

SAM Audio is available through the Segment Anything Playground, where users can test the model with sample assets or upload their own files. Meta has also made the model available for download, signalling broader ambitions to make audio segmentation a core capability of its AI ecosystem.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

BioTechEU aims to close Europe’s biotech funding gap

The European Commission and the European Investment Bank Group have launched BioTechEU, a new initiative to mobilise €10 billion in investment for biotechnology and life sciences between 2026 and 2027.

The programme targets Europe’s biotech funding gap, seeking to strengthen global competitiveness by channelling public and private capital into health innovation, including gene therapies, mRNA treatments, personalised medicine and AI-enabled medical technologies.

BioTechEU will operate under the EIB Group’s TechEU framework and draw on instruments such as the InvestEU guarantee. The initiative aligns with broader EU efforts to retain strategic health innovation within Europe and reduce reliance on external markets.

European Health Commissioner Olivér Várhelyi said under-investment continues to constrain biotech startups, adding that the European Commission sees BioTechEU as a way to help promising treatments scale and reach patients more efficiently across the EU.

EIB President Nadia Calviño said Europe has strong scientific talent and ideas, but deeper capital markets are needed. She described BioTechEU as a catalyst for enabling EU-based biotech companies to grow and compete globally.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

UN member states adopt WSIS+20 outcome document

The WSIS+20 review process – dedicated to reviewing progress made in the implementation of outcomes of the World Summit on the Information Society, 20 years after their adoption – finalised in New York, with the adoption of an outcome document at the end of a dedicated high-level meeting of the General Assembly. Following several months of consultations and negotiations, the document takes stock of progress made towards the WSIS vision of a people-centred, inclusive, and development-oriented information society, while identifying areas where further efforts and strengthened cooperation remain necessary.

The outcome document contains several provisions on the WSIS architecture, reaffirming existing mechanisms and introducing some adjustments aimed at strengthening implementation, coherence, and follow-up. One significant decision made as part of the WSIS+20 process concerns the Internet Governance Forum (IGF). Established in 2005 with a time-bound mandate that was renewed in 2010 and 2015, the IGF is now made a permanent forum of the United Nations. This decision reflects broad support among member states and was widely welcomed by non-governmental stakeholders as well.

In addition to making the IGF permanent, the outcome document introduces several measures intended to enhance its functioning and impact. The IGF is called upon to improve its work modalities and to broaden participation, particularly by governments and stakeholders from developing countries and underrepresented communities. It is invited to reinforce intersessional work, strengthen support for national and regional IGF initiatives, and apply innovative, inclusive, transparent, and agile collaboration methods.

The document also calls for the strengthening of the IGF Secretariat and requests the Secretary-General to submit a proposal to the General Assembly to ensure sustainable funding for the Forum. The IGF is further requested to report annually on progress to the Commission on Science and Technology for Development (CSTD) and to report its outcomes to relevant UN entities and processes, with a call for the UN Group on the Information Society (UNGIS), action line facilitators, the WSIS Forum, and other relevant bodies to take IGF outcomes into account in their work.

One interesting point negotiated among member states concerned the establishment of a governmental segment at the IGF. Some member states viewed this track as an important element towards fostering more dialogue among governments on digital governance issues (for some, it was also a response to the call for enhanced cooperation in the Tunis Agenda).

Others were concerned that such a segment would shift away from the IGF’s multistakeholder nature (despite the fact that the IGF, at the moment, has dedicated tracks for various groups such as parliamentarians). The final text is meant to be a compromise: The Forum is called upon to work on ‘establishing and facilitating a dialogue among Governments with the participation of all stakeholders’. 

Beyond the IGF, member states agreed that the WSIS Forum should continue to be held on an annual basis and invited the UNGIS to enhance its agility, efficiency, and effectiveness, as well as to expand its membership.

Additional provisions aim to strengthen coherence across UN digital processes. Action line facilitators are requested to develop targeted implementation roadmaps linking WSIS action lines with relevant Sustainable Development Goal targets and Global Digital Compact (GDC) commitments. Furthermore, UNGIS is requested to prepare a joint implementation roadmap to strengthen coherence between WSIS and the GDC, to be presented to CSTD in 2026. The Secretary-General is requested to submit a biennial report on WSIS implementation progress, to be considered by CSTD and ECOSOC, and the General Assembly is requested to convene a further high-level review of WSIS outcomes in 2035.

Throughout the WSIS+20 process, many discussions focused on the interplay between WSIS and GDC processes and the need to avoid duplication and enhance synergies. This is recognised in the outcome document, and several provisions – in particular those related to the implementation roadmaps, coupled with other elements describing roles for the UN Secretary-General, CSTD, the Economic and Social Council, and the General Assembly – offer important pathways in this regard. Moving forward, the key will be in how these provisions are implemented.

Substantively, the outcome document places the closure of digital divides at the core of the WSIS+20 agenda. It addresses multiple and intersecting dimensions of digital exclusion, including accessibility and equal access, inclusion of people in vulnerable situations and those in underserved, rural, and remote areas, affordability and quality of connectivity, multilingualism, cultural diversity, and the commitment to connect all schools to the Internet. The document emphasises that digital inclusion requires more than connectivity alone and must be supported by skills development, enabling environments, and respect for human rights.

The document also underscores the importance of fostering an open, fair, and non-discriminatory environment for digital development, including in the context of the digital economy. It highlights the need for predictable and transparent policy, legal, and regulatory frameworks, calls for technical assistance and technology transfer to developing countries on mutually agreed terms, and reiterates the call for states to refrain from unilateral economic measures not in line with international law. Environmental sustainability is also covered, with commitments to leverage digital technologies for sustainability while addressing energy use, critical mineral resources, e-waste management, and the development of international standards for sustainable digital products.

Human rights and ethical considerations are reaffirmed as foundational to the information society. The outcome document reiterates that the same rights apply online and offline, commits to safeguards to prevent and address adverse human rights impacts of digital technologies, and calls on the private sector to respect human rights throughout the technology lifecycle. It addresses concerns related to violence, hate speech, discrimination, misinformation and disinformation, cyberbullying, and child sexual exploitation and abuse, while emphasising information integrity, media freedom, privacy, freedom of expression, and the need to refrain from internet shutdowns and unlawful surveillance practices.

Capacity development and financing are treated as enablers of implementation. The document highlights the need to strengthen digital skills, policy and technical expertise, and institutional capacity, including in relation to emerging technologies such as AI. It invites the International Telecommunication Union – as Secretariat of UNGIS, and working with WSIS Action Line facilitators and other group members – to establish an internal task force to assess gaps and challenges in financial mechanisms for digital development and to report recommendations to CSTD in 2027.

It also calls on the Inter-Agency Working Group on Artificial Intelligence to map existing UN capacity-building initiatives, identify gaps, and address them, including through the establishment of an AI capacity-building fellowship for government officials and research programmes. These elements were subject to substantive discussions during the negotiations, with some members supporting them as important for building more capacities in AI, and others expressing concerns over potential duplication with existing work.

The outcome document reinforces the importance of monitoring and measurement, requesting a systematic review of existing ICT indicators and methodologies; the Partnership on Measuring ICT for Development is requested to conduct this review, in cooperation with action line facilitators and with the support of the Statistical Commission, and to report to CSTD in 2027. Finally, the document reaffirms the role of CSTD, ECOSOC, and the General Assembly in overall WSIS follow-up and review. 


Read more:

AI governance talks deepen as BRICS aligns national approaches

BRICS countries are working to harmonise their approaches to AI, though it remains too early to speak of a unified AI framework for the bloc, according to Deputy Foreign Minister Sergey Ryabkov.

Speaking as Russia’s BRICS sherpa, Ryabkov said discussions are focused on aligning national positions and shared principles rather than establishing binding standards, noting that no common BRICS AI rules have yet taken shape.

He highlighted the adoption of a standalone leaders’ declaration on global AI governance at the Rio de Janeiro summit, describing it as a milestone for the organisation and a first for the grouping.

BRICS members, including Russia, view cooperation on AI as a way to manage emerging risks, build capacity and help narrow the digital divide, particularly for developing countries.

Ryabkov added that the group supports a central coordinating role for the United Nations, with AI governance anchored in national legislation, respect for sovereignty, data protection and human rights.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!