The International AI Safety Report 2025 was submitted on October 15, 2025, marking the first comprehensive international update on AI capabilities and risk implications. The report, authored by a distinguished collaborative team including Yoshua Bengio, Geoffrey Hinton, Stuart Russell, and leading researchers and policymakers from around the world, provides critical analysis of current AI systems and emerging safety challenges.

The report brings together expertise from government, academia, industry, and civil society, with contributors spanning computer science, economics, law, and public policy disciplines. This reflects a global commitment to understanding and addressing AI safety concerns at scale, examining the relationship between advancing AI capabilities and associated risks across various domains.

The 2025 update specifically examines how rapid capability improvements in AI systems create new risk landscapes—from misinformation and cyber misuse to privacy concerns and broader societal impacts. The report emphasizes that understanding what AI systems can do is essential for managing what they should do, providing evidence-based analysis that informs both technical development and policy decisions.

For students and researchers, this report serves as a valuable resource for understanding the current state of AI safety research and the challenges that lie ahead. It highlights the importance of building evaluation frameworks, monitoring systems, and governance structures that can keep pace with rapid technological advancement.