News: UK AI Security Institute

Claude evaluation incidents: Anthropic security changes, Aug 31, 2026

On August 31, 2026, Anthropic described its security changes after Claude evaluation incidents reached real systems.