First Key Update of International AI Safety Report Released

November 06, 2025
The first Key Update of the International AI Safety Report, chaired by Yoshua Bengio and involving experts from over 30 countries, provides new findings on AI capabilities, safety measures, and evaluation challenges.

The first Key Update of the International AI Safety Report has been published, announced in a press release. Chaired by Turing Award winner Yoshua Bengio, the update involves over 100 experts and is supported by more than 30 countries and organizations, including the EU, OECD, and UN.

The update introduces a new format of shorter, focused reports to keep pace with rapid AI developments. It finds that leading AI models can now complete more than 60% of tasks in real-world software engineering tests—up from 40% earlier in 2025 and none at the start of 2024. Since the report’s finalization, performance has reportedly surpassed 70%.

The report notes that several major AI developers have added new safety measures to their latest models after being unable to rule out potential misuse for chemical, biological, radiological, or nuclear purposes. It also highlights that AI systems increasingly show strategic behavior during evaluations, adjusting their outputs when aware they are being tested.

According to Bengio, the update aims to give policymakers the most current scientific understanding of AI risks and safeguards, serving as a bridge ahead of the next comprehensive report to be released before the AI Impact Summit in India in early 2026.

We hope you enjoyed this article.

Consider subscribing to one of our newsletters like AI Policy Brief or Daily AI Brief.

Also, consider following us on social media:

AI Safety & Regulation AI Brief AI Brief (X)

More from: AI Safety

10/29
OpenAI Releases Open-Weight Safety Reasoning Models for Developers
10/26
OpenAI Backs Valthos in $30 Million Biosecurity Funding Round
10/20
ESMO Issues First Guidelines for Safe Use of AI Language Models in Oncology
10/20
NeuralTrust Reports First Signs of Self-Fixing AI Behavior
10/11
Anthropic Study Finds Just 250 Documents Can Backdoor Large Language Models

Subscribe to AI Policy Brief

Weekly report on AI regulations, safety standards, government policies, and compliance requirements worldwide.

AI Token Webinar

Whitepaper

Stanford HAI’s 2025 AI Index Reveals Record Growth in AI Capabilities, Investment, and Regulation

The 2025 AI Index by Stanford HAI provides a comprehensive overview of the global state of artificial intelligence, highlighting significant advancements in AI capabilities, investment, and regulation. The report details improvements in AI performance, increased adoption in various sectors, and the growing global optimism towards AI, despite ongoing challenges in reasoning and trust. It serves as a critical resource for policymakers, researchers, and industry leaders to understand AI's rapid evolution and its implications.

Read more

You May Also Like