MythosWatch: Tracking who has access to Anthropic's Mythos AI

Anthropic's Mythos Preview AI model has experienced unauthorized access through a third-party vendor, with 51+ entities now having confirmed or reported access including government agencies (NSA, White House, Pentagon), while 40+ organizations remain undisclosed—creating significant security, compliance, and competitive intelligence risks for enterprises relying on or competing with frontier AI capabilities. The fragmented access landscape, regulatory scrutiny from CISA and international bodies, and ambiguous supply chain security controls signal that CIOs must treat cutting-edge AI vendor security with the same rigor as critical infrastructure, given the geopolitical and operational implications of who controls access to state-of-the-art models.