A new International AI Safety Report has been published at internationalaisafetyreport.org, offering a comprehensive assessment of the capabilities, risks, and safeguards associated with advanced artificial intelligence systems. The report represents a major international scientific effort, drawing on contributions from more than 100 independent experts nominated by governments and international organisations.
Led by renowned AI researcher Yoshua Bengio, the report highlights rapid advances in general-purpose AI alongside escalating risks, including misuse of deepfakes, large-scale disinformation, and insufficient oversight of increasingly autonomous systems. A central finding is that AI development is outpacing the ability of institutions to test, regulate, and govern these technologies effectively.
Positioned as a neutral scientific reference rather than a policy manifesto, the report aims to provide governments and industry with a shared evidence base to support international coordination on AI safety. The full report and summaries for policymakers are publicly available on the official website.
