An intelligence explosion would compress years of progress into months and create urgent challenges for oversight, control, and governance.
We construct an occupation-level adaptive capacity index that measures a set of worker characteristics relevant for navigating...
Embedded assessments can enable deeper and more flexible assessments of risks that depend on internal systems and practices.
Anthropic’s Responsible Scaling Policy (RSP) – its framework for managing catastrophic risks – has undergone its most significant update yet.
GovAI's Annual Report 2025.
Private actors are currently well-positioned to provide most of the systems and protocols needed to make AI agents useful and safe. However, governments may still have important roles to play.
.webp)
