Anthropic's Mythos mess just keeps getting more complicated
Anthropic's internal controversy over its Mythos project grows increasingly complex, with new allegations and complications emerging beyond the initial whistleblower reports, deepening the scrutiny on the AI company's practices and culture.
Background
- Anthropic is an AI safety company founded by ex-OpenAI employees, best known for building the Claude family of large language models.
- "Mythos" (previously "Mythic") is a mysterious, reportedly secret AI model or project inside Anthropic that has been the subject of leaks, internal board drama, and public confusion since early 2026.
- The article updates a running saga: staff disagreements over whether Mythos is too powerful/dangerous to deploy, conflicting statements from Anthropic leadership, and rumors of a rift between the "effective altruism" safety faction and the "move fast" commercial faction within the company.
- This matters because Anthropic positions itself as the responsible, safety-first alternative to OpenAI and Google — so internal chaos over a secret model undercuts that brand and raises questions about how AI companies actually manage risk behind closed doors.