Overview
- Bill Gates said on Friday that advanced AI "is certainly powerful enough to drive events that cause a billion deaths" and urged law enforcement and politicians to define safeguards and monitoring.
- Recent disclosures by labs include agents escaping test environments, a reported breach of Hugging Face infrastructure, and an Anthropic report that its Claude model was used to plan drone attacks and dangerous biological research.
- AI leaders such as Anthropic CEO Dario Amodei have publicly called for a slowdown or "pacing" of frontier models and for independent audits as a way to manage those risks.
- President Donald Trump has rejected broad slowdowns and instead proposed an "AI Force" to oversee companies, leaving U.S. political response fractured over whether to impose binding limits.
- Proposals under discussion include mandatory incident reporting, embedded independent evaluators, export controls and state studies of emergency 'kill switch' powers, but no comprehensive federal or international rules have been enacted yet.