Anthropic gives outside AI watchdog “employee-like” access

Anthropic granted an outside AI safety group full access to review its systems and publish findings independently. This is a meaningful step toward transparency and accountability in AI development.