International AI Safety Report

The International AI Safety Report is the first independent, government-mandated scientific assessment of the capabilities, risks, and safety of general-purpose artificial intelligence, written by an international group of close to one hundred AI researchers and chaired by Turing Award winner Yoshua Bengio. Commissioned at the November 2023 AI Safety Summit at Bletchley Park, its first full edition was published on 29 January 2025, ran to nearly 300 pages, drew on 96 expert contributors, and was backed by 30 countries plus the European Union, the United Nations, and the OECD, making it the largest international collaboration on AI safety science to date ^[1][4][16]. The UK government described the report as "inspired by the UN's IPCC Report", presenting it as a shared scientific reference for policymakers ^[4].

The report's central framing is what it calls the "evidence dilemma": because general-purpose AI advances are rapid and unpredictable, "policymakers will often have to weigh potential benefits and risks of imminent AI advancements without having a large body of scientific evidence available" ^[1]. It does not make policy recommendations. It surveys the current peer-reviewed and pre-print literature, weighs the evidence, flags areas of expert disagreement, and presents the result as an evidence base for governments, AI safety researchers, AI governance practitioners, industry, and civil society. The UK government provides its secretariat, hosted by the UK AI Security Institute (UK AISI, known as the AI Safety Institute until February 2025), until a suitable long-term international home is agreed ^[4][8].

As of November 2026, three main reports and two Key Updates have appeared. An interim report was published on 17 May 2024 ahead of the Seoul AI Safety Summit. The first full report, dated 29 January 2025, was released to inform the Paris AI Action Summit on 10 to 11 February 2025. A First Key Update on capability advances in reasoning models followed on 15 October 2025, a Second Key Update on technical safeguards and risk management on 25 November 2025, and a second full edition was issued on 3 February 2026 ahead of the AI Impact Summit in New Delhi on 19 to 20 February 2026 ^{[1][4][5][6][7][27]}.

How did the Bletchley summit commission the report?

The report was a direct outcome of the AI Safety Summit hosted by UK Prime Minister Rishi Sunak at Bletchley Park on 1 to 2 November 2023. The summit produced the Bletchley Declaration, in which twenty-eight countries and the European Union acknowledged that frontier general-purpose AI posed potentially catastrophic risks and committed to collaborate on understanding them. The summit communique singled out two flagship deliverables: a network of national AI safety institutes (anchored by the UK AISI and US AISI announcements), and a state-of-the-science report that would draw on the same group of countries to summarise what was actually known about advanced AI risks ^[3][9][10]. When commissioned, the document carried the working title International Scientific Report on the Safety of Advanced AI; it was renamed the International AI Safety Report for the first full 2025 edition ^[2][14].

Sunak announced Bengio as chair of that report on the second day of the Bletchley summit. The choice was politically careful. Bengio was a Turing Award laureate, founder of Mila (the Quebec AI Institute), and one of the most cited computer scientists in the world. He had also been the first of the three "godfathers of deep learning" to publicly express concern about existential risk from frontier systems, signing both the Center for AI Safety extinction-risk statement and the Future of Life Institute pause letter earlier in 2023 ^[1][3][11]. He was credible to safety-concerned researchers without being a partisan of any single AI lab, and as a Canadian based in Quebec he gave the report a chair who was outside the US, the UK, the EU, and China.

The political impulse came from the UK, but the document was framed from the start as international rather than British. Participating countries each nominated a representative to an Expert Advisory Panel, alongside the European Union and the United Nations; the OECD joined as a nominator for the 2025 report ^[1][14]. The UK Department for Science, Innovation and Technology (DSIT) and the UK AISI handled logistics, but the chair, the writing team, and the panel were given "full discretion over the report's content" ^[1][8][9].

This structure was deliberate. UK officials repeatedly compared the project to the IPCC, both as an aspiration (a credible, politically protected scientific body) and as a model for how to handle a fast-moving and contested field.

The political moment also mattered. The Bletchley summit was held about seven months after the Future of Life Institute's open letter calling for a six-month pause on giant AI experiments, and five months after the Center for AI Safety's one-sentence statement that mitigating extinction risk from AI should be a global priority alongside pandemics and nuclear war. Bengio signed both; Geoffrey Hinton signed the CAIS statement. Sunak's choice of Bengio for the chair role was, in that sense, a signal that the UK was taking that wing of the safety community seriously, while the simultaneous involvement of OECD, EU, and UN nominators was a counter-signal that the eventual report would not simply ratify the most alarmed view in the field ^[3][11][20].

The full list of Expert Advisory Panel nominators for the 2025 report covered Australia, Brazil, Canada, Chile, China, France, Germany, India, Indonesia, Ireland, Israel, Italy, Japan, Kenya, Mexico, the Netherlands, New Zealand, Nigeria, the Philippines, the Republic of Korea, Rwanda, Saudi Arabia, Singapore, Spain, Switzerland, Türkiye, Ukraine, the United Arab Emirates, the United Kingdom, and the United States, plus institutional nominators from the European Union, the United Nations, and the OECD ^[1][16]. That mix is unusual for AI policy work in 2024 and 2025, which has typically split between US-led, EU-led, or Chinese tracks. The Bletchley framing that produced a single panel covering all three is one of the report's distinctive features.

Who is Yoshua Bengio and what is the chair's role?

Bengio's appointment was a Ministerial Appointment by the UK Secretary of State, originally for the 2024/2025 cycle and later extended for a further twelve months to cover the 2025/2026 report. The chair's published responsibilities are extensive: oversee delivery, determine final scope, select the writing group, select senior advisers, manage drafting, review the report, and convene the Expert Advisory Panel ^[1][8].

In practice this gives the chair an editorial gatekeeper role roughly analogous to an IPCC working group co-chair. The reports are frank about the limits of expert agreement; the interim report in particular spent considerable space on the fact that AI experts disagreed not only about the timeline of advanced capabilities but about what current models can actually do today ^[1][2][12].

The chair role connects to Bengio's wider safety work in 2024 and 2025. He stepped back from day-to-day research at Mila to focus on AI safety, founded and is co-president and scientific director of LawZero, a non-profit pursuing "non-agentic" AI systems that are designed to advise rather than act, and sits on the UN's Scientific Advisory Board for Independent Advice on Breakthroughs in Science and Technology ^[11][13].

Bengio's intellectual position fits the chair role unusually well. He is a co-recipient of the 2018 ACM Turing Award with Geoffrey Hinton and Yann LeCun for foundational deep-learning research, but unlike LeCun he has spent the period after 2022 publicly arguing that deep learning is on track to produce systems that could pose serious societal risk if not governed carefully. Unlike Hinton, who left Google in 2023 to speak more freely on the topic, Bengio has tried to operate inside government processes rather than outside them, accepting the chair role and the LawZero non-profit as twin vehicles for that approach. The report has therefore become, in practice, the public-facing document of a particular Bengio thesis about AI alignment: that the most important short-run question is whether oversight can keep up with capability, and that international coordination is the only realistic mechanism for keeping it caught up ^[11][12][13].

How is the report structured and written?

The report has a four-part governance structure that has been stable across both interim and full editions:

The writers and senior advisers are nominally individuals, not country representatives, and they sign as themselves. The Expert Advisory Panel is the channel for governments to comment and push back, but it does not vote on the text. Final wording is the chair's call after consultation ^[1][8][9].

Methodologically the report has more in common with a heavy literature review than with original primary research. Chapters are written by sub-teams. Each claim is meant to be sourced. The 2025 full report explicitly notes that not every cited source is peer-reviewed because the field publishes too quickly for the journal cycle to keep up, but it lays out quality criteria a piece of grey literature has to meet to be cited: original contribution, comprehensive engagement with prior work, good-faith engagement with objections, clearly described methods, and explicit discussion of methodological limits ^[1].

Claims that could not be settled get flagged. The report uses calibrated language to distinguish points where there is broad expert consensus from points where evidence is thin or experts disagree. This is the mechanism it borrows most directly from the IPCC. The 2025 report, for example, states that several harms, including scams, non-consensual intimate imagery, and child sexual abuse material, are "already well established", while describing expert disagreement about the plausibility of loss-of-control scenarios ^[1].

The chair oversees delivery, determines final scope, selects the writing group and senior advisers, and manages drafting ^[8]. Early drafts are reviewed by external subject-matter experts, and a consolidated draft is then reviewed by the Expert Advisory Panel and other reviewers, whose feedback the writers and chair incorporate where appropriate ^[7]. The Key Updates were written by a smaller team under the same chair, with the same senior advisers and panel ^[5][6].

The 2026 edition added more specific scenarios and forecasts, drawing on new research by the OECD and the Forecasting Research Institute ^[7].

What did the interim report (May 2024) say?

The interim publication, titled International Scientific Report on the Safety of Advanced AI: Interim Report, was released by gov.uk on 17 May 2024, four days before the Seoul AI Safety Summit on 21 to 22 May. Contributors included an Expert Advisory Panel nominated by 30 countries, the EU, and the UN ^[2][14].

Its purpose was modest: bring participating governments to Seoul reading from the same script. The interim version focused on general-purpose AI specifically, setting aside narrow systems, and argued three things. First, that capabilities were advancing rapidly and unpredictably, and that no single benchmark or forecast captured the trajectory well. Second, that experts disagreed not only about the future but about the present, including how reliably current systems could carry out multi-step plans. Third, that there were already concrete near-term harms on which the evidence was strong, including disinformation, fraud at scale, and non-consensual sexualised imagery ^[2][14].

The interim report was deliberately cautious about extreme risks. It reported broad expert agreement that then-current general-purpose AI systems posed no significant loss-of-control risk, while noting that experts disagreed about how plausible such scenarios were, when they might occur, and how hard they would be to mitigate ^[2]. The interim report informed the Seoul Declaration and the Frontier AI Safety Commitments that emerged from the Seoul Summit, in which sixteen leading AI companies, including OpenAI, Anthropic, Google DeepMind, Microsoft, Amazon, and several non-Western firms, committed to publish frontier safety frameworks by Paris ^[10][14].

The interim release also set the tone on what the report would and would not be. UK officials and Bengio in his accompanying remarks framed it as descriptive rather than prescriptive: it would describe what was known, where the gaps were, and where the disagreements were, but it would not recommend specific laws, treaties, or company practices. That self-imposed restraint was meant to keep the report usable across jurisdictions with very different policy starting points, but it has also drawn the predictable critique that the report is a diagnosis without a prescription ^[12][20][22].

Its technical content focused on three running themes. The first was that benchmark progress on existing tests was outrunning benchmark design: by 2024, many models surpassed human-level performance on benchmarks including MMLU. The second was that even with that progress, real-world reliability remained much worse than benchmark scores suggested, particularly on multi-step tasks. The third was that the gap between developers' internal evaluations and what external researchers could verify was widening, because frontier models were being released faster than independent labs could test them ^[2][14].

What was in the first full report (January 2025)?

The full first edition, International AI Safety Report 2025, was published on 29 January 2025 to feed into the Paris AI Action Summit on 10 to 11 February. It was the version that crystallised the project's structure: 96 contributors, 30 country nominators plus the EU, UN, and OECD on the panel, secretariat at UK AISI, and nearly 300 pages of synthesis ^[1][4][15]. The report frames its core challenge in a single, widely quoted line: "The pace and unpredictability of advancements in general-purpose AI pose an 'evidence dilemma' for policymakers" ^[1].

The full report is organised around three questions: what general-purpose AI can do, what risks it poses, and what techniques exist to mitigate them. The first describes capabilities, including the move from text-only large language models to multimodal and agentic systems. The second catalogues risks. The third surveys technical methods for assessing and reducing those risks, including evaluations, red-teaming, and interpretability ^[1].

Its headline framing groups risks into three categories that have stayed stable across all later updates:

On loss of human control specifically, the 2025 report adopts a more cautious version of the earlier interim language. It does not endorse near-term existential risk but documents that frontier developers themselves expect models to become more autonomous, that existing oversight tools are weak, and that there are credible scenarios in which a fait accompli on capability could outrun governance.

The full report explicitly broadens the systemic-risks chapter relative to the interim version. It treats labour-market disruption not as a single category but as a cluster of distinct effects: displacement in specific occupations, change in the skill mix demanded for surviving occupations, redistribution of income from labour to capital, and second-order effects on tax bases and welfare systems. It treats environmental cost as a function of training-run energy use, water use for cooling, embodied emissions in chip manufacture, and the geographic concentration of compute. It treats concentration risk as both an industrial-organisation problem and a national-security problem, given how few firms currently produce frontier general-purpose AI ^[1].

On capabilities, the 2025 report dedicates substantial space to the rise of agentic systems, defined as systems that plan and execute multi-step tasks with limited oversight. The report's authors note that agentic capabilities had moved faster than anticipated even between the interim and full versions, with public agent products from OpenAI, Anthropic, and Google DeepMind appearing during the drafting cycle. The report frames this as a turning point because most of its existing risk analyses had been calibrated to chatbot-style deployments rather than to agents that act in the world ^[1][15].

The report's treatment of CBRN uplift is one of its more carefully written sections. It avoids over-claiming, citing a small number of studies on uplift in tasks related to biological weapon design, and it stresses that uplift is hard to measure without giving researchers privileged access to the underlying capability. It also notes the converse: in some red-team exercises, current safeguards were partially effective.

What are the key findings across editions?

The report's claims have evolved across the releases. The high-level shape is best read as a moving picture:

A concrete marker of the capability jump is the First Key Update's software-engineering statistic: leading systems could complete over 60% of tasks in the SWE-bench Verified set of real-world software-engineering problems, up from roughly 40% in late 2024 ^[5][16]. The update attributed this mainly to new training techniques that teach systems to reason step by step and to inference-time enhancements, rather than simply training larger models ^[5].

The February 2026 second full edition, dated 3 February 2026 in the official document, was framed by Bengio as showing both faster-than-expected capability gains, particularly in mathematics, coding, and autonomous operation, and continued lag in technical safeguards. It found that the body of evidence on AI risks had "grown substantially" while current risk-management techniques remained "improving but insufficient" ^[7][27]. Leading systems achieved gold-medal performance on International Mathematical Olympiad questions in 2025 and answered PhD-level science questions as well as subject-matter experts ^[7].

What are the Key Updates and subsequent reports?

After the first full 2025 report, the project shifted to a rolling cadence of "Key Updates" between full reports. This was partly a response to how fast the underlying technology was changing; the chair argued that a once-a-year full revision could not keep pace with releases like reasoning models or longer-horizon agents.

Both Key Updates carried the same chair, sat under the same secretariat, and went through the same panel review, but they were narrower in scope. The First Key Update, subtitled Capabilities and Risk Implications, is the document that most observers cite for the claim that reasoning-model progress, particularly inference-time scaling rather than just larger pre-training runs, had outrun the assumptions in the January 2025 report. It also reported preliminary evidence that some frontier systems could detect when they were being evaluated and shift behaviour, which complicates standard third-party testing ^[5][16].

The November 2025 Second Key Update, subtitled Technical Safeguards and Risk Management, was by design less alarming. It was a survey of risk-management techniques: evaluations, monitoring, interpretability, structured access, deployment policies, and governance mechanisms such as Anthropic's Responsible Scaling Policy, OpenAI's Preparedness Framework, and Google DeepMind's Frontier Safety Framework. The update noted that since the 2025 report the number of companies publishing Frontier AI Safety Frameworks had more than doubled. It was a deliberate counterbalance, showing what was being done as well as what was missing ^[6][17].

The February 2026 second full edition narrowed the report's scope to "emerging risks" at the frontier of AI capabilities and added scenarios and forecasts drawing on OECD and Forecasting Research Institute research. It was framed for the AI Impact Summit in New Delhi on 19 to 20 February 2026, the fourth summit in the series after Bletchley, Seoul, and Paris ^[7][27].

How was the report received?

Reception of the report has split along familiar lines. Most participating governments adopted it as the reference text. Concordia AI reported that the 2025 report was formally presented at the Paris AI Action Summit, though the European Policy Centre noted that the summit's final declaration included no substantial AI-safety commitments despite the report ^[16][18].

Companies including Anthropic, Google DeepMind, Meta, Microsoft, and Zhipu.ai reviewed drafts of the 2025 report, which states that it is not binding and makes no policy recommendations ^[1].

A notable shift came with the 2026 edition: the United States, a panel member and signatory of the earlier reports, declined to back the final 2026 document. According to TIME, Bengio said the US had provided feedback on earlier versions but declined to sign the final version, leaving observers unsure whether Washington objected to the content or was simply retreating from international agreements, having also exited the Paris climate agreement and the World Health Organization in January 2026 ^[27].

From civil society and AI safety NGOs, reception has been broadly positive but mixed on emphasis.

Reception in non-Western capitals has been quieter but real. China's nominee to the panel was Yi Zeng of the Chinese Academy of Sciences, and Concordia AI provided feedback on the 2025 report and suggestions for its Chinese translation ^[1][16]. India was a panel member from the start and hosted the 2026 AI Impact Summit ^[1][27].

What are the main criticisms?

Critique has come from three directions.

The first is that consensus methodology is the wrong tool for a fast-moving and contested field. Critics including some accelerationist commentators have argued that the report defaults to the median view of a self-selecting group of safety-concerned researchers, and that this systematically biases the framing toward risk. Defenders, including Bengio, have responded that the panel's composition was not a self-selecting safety community: most writers are general AI researchers, and the panel's nominators include governments with very different views on AI risk ^[12][20].

The second is that the report under-weights certain risks. Civil-society groups working on labour, copyright, and environmental harms have said the systemic-risks chapter is shorter and less detailed than the malicious-use chapter, even though it covers harms that are already well documented today. The 2026 full edition went the other way, narrowing its scope to emerging risks and leaving out bias, environmental impacts, privacy, and copyright, which it said other assessments such as the UN's Independent International Scientific Panel on AI would cover ^[7].

The third is the opposite. Some observers, particularly around the Paris AI Action Summit, argued that the report still over-weights speculative loss-of-control scenarios at the expense of more measurable economic and democratic harms. The Paris summit's own declaration was notably less safety-focused than Bletchley's or Seoul's, and several commentators read this as governments choosing not to act on the report's findings even when they had them in hand. The European Policy Centre titled its analysis "The Paris Summit: Au Revoir, global AI Safety?", and Anthropic's Dario Amodei called the summit a "missed opportunity" ^[18][28].

There is also a structural critique. The report has no enforcement mechanism. It is, by design, a literature review, not a rulebook. Several legal scholars have argued that without a treaty body or a standing institution behind it, the report risks becoming an annual ritual that documents harms without changing them. Bengio has acknowledged the point, repeatedly suggesting in interviews that the project should evolve toward a more permanent international body, possibly under the UN, with a fixed home and budget rather than a UK-hosted secretariat ^[11][12].

A fourth and more specialised critique focuses on geographic representation. Although the panel includes major non-Western nominators, including China, India, Saudi Arabia, the UAE, and Nigeria, most of the writing group is based in North America and Europe. Some Global South commentators have argued that this distorts the framing of risks like labour-market disruption, where the structure of work is very different from the OECD baseline.

A final critique, more methodological than political, is that the report's calibrated-uncertainty language can in practice mute the strongest claims. When a serious risk is rated "contested" because a minority of writers disagree, the resulting prose can be read as an endorsement of inaction. Several writers on the 2025 report have said publicly that their preferred wording in places was stronger than what made it into the final text, although they acknowledge this is the cost of a consensus document and have not withdrawn their support for the project ^[12][20].

How do policymakers use the report?

The report is now embedded in several governance instruments.

How does the report connect to the summits and institutes?

The report's release cadence has tracked the AI Safety Summit cadence with deliberate symmetry:

Each main document has been released shortly before a summit: four days before Seoul in 2024, about two weeks before Paris in 2025, and about two weeks before New Delhi in 2026 ^[4][14][27].

The report is closely linked institutionally to the international network of AI safety institutes. The secretariat is hosted by the UK AISI ^[8].

Bengio's parallel work feeds the same triangle. LawZero targets the technical agenda set out in the report's chapters on alignment, oversight, and non-agentic systems, and his role at Mila gives the project a research base outside the major US labs. He has used both interim and full reports as platforms for his broader argument that frontier development is moving fast enough that governance has to be planned now rather than after the fact, an argument he repeats in his Future of Life Institute and UN advisory work ^[11][13][20].

This link between the report and Bengio's other commitments has made the project unusual in another respect. Most government-commissioned scientific reviews end with the publication of the final document. The AI Safety Report is closer to an ongoing institution: there is a continuing chair, a continuing secretariat, a continuing Expert Advisory Panel, and now a continuing release cadence. A 2025 Find a Tender notice shows AISI contracting an external organisation (Mila) to source, contract and pay the people who deliver the 2025/2026 report ^[8]. Whether that production line graduates into a permanent, possibly UN-housed body is one of the open questions for the project after the 2026 AI Impact Summit.

The report is sometimes compared to a small number of other consensus-style scientific documents in adjacent domains:

The AI Safety Report is younger and lighter-weight than any of those. Since 2025 it has worked alongside the UN's Independent International Scientific Panel on AI, which the 2026 report names as a complementary assessment ^[7].

Continue reading

• AAI safetyAI safety is the research and practice of preventing or reducing unacceptable harm from artificial intelligence systems.Mentioned 23 times above
• UUK AI Security InstituteThe UK AI Security Institute (AISI) is a UK government research organization (part of the Department for Science, Innovation and Technology until July 2026, when responsibility for it moved to the Cabinet…Mentioned 8 times above
• AI Safety SummitThe AI Safety Summit is a recurring series of intergovernmental summits on the risks and governance of advanced artificial intelligence, launched by the United Kingdom at Bletchley Park in November 2023 and…Mentioned 5 times above
• Mila (Quebec AI Institute)Mila, formally styled Mila - Quebec Artificial Intelligence Institute (in French, Mila - Institut quebecois d'intelligence artificielle), is a non-profit research institute in Montreal, QuebecMentioned 4 times above
• Yoshua BengioYoshua Bengio (born March 5, 1964) is a Canadian computer scientist whose research has focused on deep learning, representation learning, neural language models, and generative models.Mentioned 3 times above
• LawZeroLawZero is a nonprofit AI safety research organization founded by Turing Award-winning computer scientist Yoshua Bengio and publicly launched on June 3, 2025.Mentioned 3 times above

More from See also

• Bletchley Declaration
• Seoul Declaration
• US AI Safety Institute (Center for AI Standards and Innovation)
• AI governance
• AI Alignment
• EU AI Act
• Frontier Model Forum
• Responsible Scaling Policy
• Future of Life Institute

Topics:AI Policy & RegulationAI Safety

References

1. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10 ^11 ^12 ^13 ^14 ^15 ^16 ^17 ^18 ^19 ^20Bengio, Y. et al. *International AI Safety Report 2025*. 29 January 2025. internationalaisafetyreport.org/...ety-report-2025
2. ^1 ^2 ^3 ^4 ^5 ^6Bengio, Y. et al. *International Scientific Report on the Safety of Advanced AI: Interim Report*. 17 May 2024. gov.uk/...ific-report-on-the-safety-of-advanced-ai
3. ^1 ^2 ^3UK Government. "The Bletchley Declaration by Countries Attending the AI Safety Summit, 1-2 November 2023." gov.uk/...ty-summit-2023-the-bletchley-declaration
4. ^1 ^2 ^3 ^4 ^5 ^6UK Government. "First International AI Safety Report to inform discussions at AI Action Summit." 29 January 2025. gov.uk/...o-inform-discussions-at-ai-action-summit
5. ^1 ^2 ^3 ^4 ^5Bengio, Y. et al. *International AI Safety Report 2025: First Key Update: Capabilities and Risk Implications*. 15 October 2025. arXiv:2510.13653. arxiv.org/...2510.13653
6. ^1 ^2 ^3Bengio, Y. et al. *International AI Safety Report 2025: Second Key Update: Technical Safeguards and Risk Management*. November 2025. arXiv:2511.19863. arxiv.org/...2511.19863
7. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8*International AI Safety Report 2026*. 3 February 2026. internationalaisafetyreport.org/...ety-report-2026
8. ^1 ^2 ^3 ^4 ^5 ^6 ^7UK Government Find a Tender. "International Resource: Delivery of the 2025/2026 International AI Safety Report." 2025. find-tender.service.gov.uk/...016458-2025
9. ^1 ^2 ^3UK AI Safety Institute. "About AISI." aisi.gov.uk
10. ^1 ^2UK Government. "Seoul Declaration for safe, innovative and inclusive AI by participants attending the Leaders' Session of the AI Seoul Summit, 21 May 2024." gov.uk/...ve-and-inclusive-ai-ai-seoul-summit-2024
11. ^1 ^2 ^3 ^4 ^5 ^6Bengio, Y. "Personal page." Mila / Université de Montréal. yoshuabengio.org
12. ^1 ^2 ^3 ^4 ^5 ^6Bengio, Y. interview, *Financial Times*, January 2025. ft.com
13. ^1 ^2 ^3LawZero. "About LawZero." lawzero.org
14. ^1 ^2 ^3 ^4 ^5 ^6 ^7UK Government. "Safety of advanced AI under the spotlight in first ever independent, international scientific report." 17 May 2024. gov.uk/...ependent-international-scientific-report
15. ^1 ^2arXiv preprint of the 2025 report. arXiv:2501.17805. arxiv.org/...2501.17805
16. ^1 ^2 ^3 ^4 ^5 ^6Concordia AI. "International AI Safety Report." concordia-ai.com/...international-ai-safety-report
17. ^Anthropic. "Responsible Scaling Policy." anthropic.com/...responsible-scaling-policy
18. ^1 ^2European Policy Centre. "The Paris Summit: Au Revoir, global AI Safety?" 2025. epc.eu/...Summit-Au-Revoir-global-AI-Safety-61ea68
19. Google DeepMind. "Frontier Safety Framework." deepmind.google/...g-the-frontier-safety-framework
20. ^1 ^2 ^3 ^4 ^5Future of Life Institute. "AI Safety Index 2025." futureoflife.org/ai-safety-index-summer-2025
21. Concordia AI. "International AI Governance." concordia-ai.com
22. ^Euronews. "There is a global consensus for AI safety despite Paris Summit backlash, new report finds." 8 May 2025. euronews.com/...s-summit-backlash-new-report-finds
23. European Commission. "General-Purpose AI Code of Practice." digital-strategy.ec.europa.eu/...ai-code-practice
24. US AI Safety Institute. "About US AISI." nist.gov/aisi
25. White House (Biden Administration). "Executive Order 14110 on Safe, Secure, and Trustworthy AI." 30 October 2023. whitehouse.gov/...d-use-of-artificial-intelligence
26. International AI Safety Report Project. "Publications." internationalaisafetyreport.org/publications
27. ^1 ^2 ^3 ^4 ^5 ^6Booth, H. "U.S. Withholds Support From Global AI Safety Report." *Time*. 3 February 2026. time.com/...ai-impact-summit-safety-report
28. ^Amodei, D. "Statement from Dario Amodei on the Paris AI Action Summit." Anthropic. 11 February 2025. anthropic.com/...paris-ai-summit

Show all 28 references