White House and Anthropic Jointly Build Framework to Rate AI Security Flaws

According to Politico, the White House and Anthropic are jointly developing a shared framework to assess the severity of security flaws in new AI models and guide potential government intervention. The central focus is building benchmarks to measure harm from jailbreaks—prompts that bypass safety guardrails. On Anthropic's side, public policy lead Sarah Heck and co-founder Tom Brown are said to be leading the talks.