Dario Amodei is an American machine learning researcher and the co-founder and chief executive of Anthropic, a company he helped start in 2021 after several years in senior research roles at OpenAI.
Early career
Amodei holds a PhD in biophysics from Princeton University. Before moving fully into machine learning, he worked in Andrew Ng's AI team at Baidu and then at Google Brain. He joined OpenAI in 2016 and became vice president of research, where he was involved in the development of GPT-2 and GPT-3 and in OpenAI's early safety and policy discussions, including the staged, cautious release of GPT-2 in 2019.
Founding Anthropic
In 2021, Amodei left OpenAI along with a group of colleagues, including his sister Daniela Amodei, Tom Brown, Jared Kaplan, and Chris Olah, to found Anthropic. The departure was widely reported at the time as stemming from disagreements over OpenAI's direction following its 2019 restructuring into a capped-profit company and its deepening partnership with Microsoft. Anthropic was built around a thesis Amodei has described repeatedly in public writing and interviews: that the safest path is for safety-focused researchers to build frontier-capability models themselves, rather than cede that ground to labs with less safety emphasis, so they can study and shape the most capable systems directly.
Public advocacy
As Anthropic's CEO, Amodei has become one of the most visible voices arguing simultaneously for continued aggressive scaling of AI capability and for substantial caution and regulation around it. Anthropic's Responsible Scaling Policy (see Responsible scaling policies) formalizes capability thresholds meant to trigger additional safety measures. In 2024, Amodei published the essay "Machines of Loving Grace," which argued that powerful AI systems could plausibly accelerate scientific and medical progress dramatically within a decade if development goes well, a notably optimistic counterpoint to his frequent warnings elsewhere about catastrophic and existential risks from advanced AI (see AI safety). He has testified before the US Congress and engaged with policymakers on AI governance.
Leadership of Anthropic
Under Amodei, Anthropic developed the Claude model family and the Constitutional AI training method, and it raised funding from investors including Google and Amazon while remaining smaller than OpenAI by most usage and revenue measures through the mid-2020s. He is frequently profiled alongside Sam Altman as one of the central figures shaping the direction of the broader generative AI boom.