Dario Amodei is the chief executive and co-founder of Anthropic, the company behind the Claude model family. He started it in 2021 with his sister Daniela and a group of colleagues from OpenAI, where he had been vice president of research. The founding pitch has not changed since: frontier AI built with safety as the engineering constraint, not the press release.
The Mythos years
Amodei spent 2025 and 2026 steering Anthropic through its most consequential program: Mythos, the internal frontier tier that surfaced through a March 2025 leak and then through Project Glasswing, the controlled-access security program that grew from 11 partners to roughly 150 organizations. The full timeline is reconstructed in our Mythos chronicle .
On June 9, 2026 the program produced its first public product: Claude Fable 5, a Mythos-class model with consumer-grade safeguards. Amodei's bet is that the safety layer, not the capability ceiling, is what makes the release defensible. We unpack that mechanic in our guardrails analysis .
In 2026
Days before the launch, Anthropic filed confidentially for an initial public offering, capping a year that included a $65 billion Series H. Where the company goes from here is the question every competitor is now pricing in. Background: Anthropic's site and his Wikipedia profile .