Claude Mythos Unauthorized Access Story Resurfaces: TechTimes Reports Anthropic's 'Most Dangerous Model' Was Breached on Day One via Third-Party Contractor Credentials

TechTimes·high signal

A trending TechTimes piece (May 17) resurfaces the Claude Mythos unauthorized access incident, detailing how a private forum group accessed the restricted frontier model on its April 8 launch day by combining credentials from a third-party Anthropic model evaluator with Mercor data breach information to guess the model's Vertex AI endpoint format. Anthropic confirmed the investigation but stated no impact beyond the vendor environment. Claude Mythos Preview remains restricted to Project Glasswing participants working on critical infrastructure — Anthropic has stated it does not plan to make Mythos-class models generally available due to their autonomous vulnerability discovery capabilities.