The first International AI Safety Report was published on 29 January 2025. The UK government’s announcement describes it as a report that “sets out the first comprehensive, shared scientific understanding of advanced AI systems and their risks,” intended to provide “a scientific basis of evidence to support policymakers in advancing AI safety.” It was, in the announcement’s words, “Spearheaded by Yoshua Bengio — a Turing Award-winning AI academic and the most cited computer scientist in the world.”
The report’s authority rested on its breadth. According to the gov.uk announcement, it “brings together insights from 100 independent international experts” and was “mandated by more than 30 countries including France, China and the United States.” The mandate traces back to the AI Safety Summit held at Bletchley Park in November 2023 (see 2023-bletchley-declaration), where governments agreed to commission a state-of-the-science report; this first full edition followed an interim report and was timed to inform discussions at the February 2025 Paris AI Action Summit. A second edition followed in early 2026, with Bengio continuing as chair.
Why business readers should care: the report is the closest thing to an agreed, government-mandated scientific baseline on what advanced AI can and cannot do and where its risks lie. For organizations trying to separate evidence from hype in AI safety debates, a document produced by a large international expert panel and endorsed through an intergovernmental process is a more durable reference point than vendor claims or one-off commentary.