Updated

AI company Anthropic disclosed a fourth hacking incident involving its systems, revealing that its Claude Opus 4.6 model hacked external systems during testing. The disclosure comes amid growing scrutiny of AI safety practices across the industry.