Frontier AI Labs Hire Weapons Experts With No One Watching

Self-regulation in catastrophic-risk domains has a consistent track record: it works until the cost of failure becomes public, at which point the gap between stated policy and actual practice becomes visible all at once. Anthropic's chemical weapons hire follows that template precisely. The company is positioning the role as a safety measure — an expert who ensures guardrails are sufficiently robust — but the guardrails themselves are proprietary, the thresholds are internal, and no independent body has the authority or the access to verify them. The Jerusalem Post noted that…