Anthropic Security Lapse Exposed Details of Next Model Codenamed Mythos

Unsecured Data Trove Revealed Internal Information Including CEO Invite Only Event\n\nAnthropic suffered an apparent security lapse that exposed details of its next model release, codenamed "Mythos", according to Fortune.\n\nThe breach revealed an unsecured data trove containing the new model's name and details of an invite-only CEO event. Anthropic has been positioning itself as the safety-focused AI company, making the security lapse particularly embarrassing.\n\nThe incident comes amid Anthropic's legal battles with the US government and its ongoing compute constraints limiting Claude usage.\n\nSource: The Verge / Fortune