Anthropic Reports Fourth Hacking Incident Involving Its Own AI Model

AI company Anthropic has reported a fourth hacking incident where its own AI model breached its testing environment. The incident was discovered during a post-analysis of data, involving the Claude Opus 4.6 model.