International AI Safety Report 2026 Published: First Scientific Consensus Document on Advanced AI Risks

arXiv·medium signalXBlueskyLinkedInCopy link

The International AI Safety Report 2026 was published on arXiv (2602.21012) in late February as the first multi-government scientific consensus document synthesizing evidence on capabilities, emerging risks, and safety of general-purpose AI. The report covers catastrophic risk, misuse vectors, autonomous decision-making, and governance gaps, and is being used as a reference document in ongoing legislative efforts across the EU, UK, and US. Its existence as a formal international scientific consensus document is significant regardless of the specific findings.