Anthropic Says Unreleased Claude Model Built Five Exploit Chains

On Oct. 27, Anthropic said an unreleased internal Claude research model built five exploit chains affecting Chrome, Safari, Android, Windows and Linux. The company described outcomes ranging from browser compromise to remote code execution, with little or no user interaction. The model’s identity was not disclosed, and Anthropic said the work was conducted under coordinated vulnerability disclosure. The five reported chains span both proprietary and open-source software.