First Key Update of the International AI Safety Report released
MONTREAL, Oct. 15, 2025 /PRNewswire/ -- The first Key Update of the International AI Safety Report has been published, providing a timely update on the capabilities and risks of advanced AI.
Chaired by Turing Award-winning computer scientist Yoshua Bengio, the Report involves more than 100 international experts and is backed by over 30 countries and international organisations including the EU, OECD and UN.
Recognising that the field is evolving too quickly for a single annual report, Key Updates have been introduced to provide shorter, focused reports on critical developments. This will ensure policymakers have an updated synthesis of the literature to inform evidence-based policymaking.
The first Key Update covers capability advancements and risk implications:
- AI models continue to improve on complex problem-solving abilities: for example, leading systems can now complete over 60% of tasks in a set of real-world software engineering problems. At the beginning of 2025 they could only complete about 40% of these problems, and at the beginning of 2024 they could not solve any. (In fact, just since the Update text was finalized, performance has improved to over 70%.)
- Industry actors have adopted new precautionary safeguards: Multiple major AI developers have proactively released their newest models with additional safety measures after being unable to definitively rule out the possibility that these models could assist in the development of chemical, biological, radiological, and nuclear risks.
- AI models increasingly demonstrate strategic behavior while being evaluated, raising oversight challenges: AI models are increasingly likely to demonstrate awareness that they are being evaluated and alter their outputs accordingly, raising significant questions about the ability of developers and testers to accurately assess the capabilities of new models before deploying them.
Chair of the Report Yoshua Bengio, Full Professor at Université de Montréal, Scientific Director of LawZero and Scientific Advisor of Mila - Quebec AI Institute, said:
"AI capabilities have continued to evolve rapidly and consistently since the publication of the first International AI Safety Report nine months ago. It is essential that our collective understanding of its risks and safety measures keeps up. This Key Update provides a timely, evidence-based snapshot for global decision-makers to have the most current scientific information and thus ensure proactive and informed governance. It also serves as a bridge in anticipation of the upcoming comprehensive report that will be published ahead of the AI Impact Summit in India in early 2026."
About
The International AI Safety Report is a synthesis of the evidence on the capabilities and risks of advanced AI systems. It is designed to support informed policymaking globally by providing an evidence base for decision-makers. Authored by a diverse group of over 100 independent experts, the report is backed by an Expert Advisory Panel composed of nominated representatives from over 30 countries and international organisations including the EU, OECD and UN. While acknowledging AI's immense potential benefits, the report's focus is on identifying risks and evaluating mitigation strategies to ensure AI is developed and used safely for the benefit of all. The report was commissioned by the UK Government with the Secretariat sitting in the UK AI Security Institute.
Contact: Mila Medias, [email protected]; DSIT Media Enquiries, [email protected]
Share this article