Dario Amodei

Dario Amodei studied at Stanford University as an undergraduate and went on to complete a PhD in physics at Princeton University, supported as a 2007 Hertz Fellow by the Fannie and John Hertz Foundation. Amodei's doctoral thesis, "Network-Scale Electrophysiology: Measuring and Understanding the Collective Behavior of Neural Circuits", earned the Hertz Foundation's thesis prize.45

After the doctorate, Amodei was a postdoctoral scholar at the Stanford University School of Medicine. Amodei has credited the Hertz Fellowship with helping, at every point in Amodei's career, to open doors and to bring exposure to new ideas.45

Amodei then moved into artificial intelligence, working at Baidu, on the Google Brain team as a deep learning researcher, and at OpenAI.54

At OpenAI, Amodei rose to vice president of research, setting the overall research direction of the organization and leading the efforts to build GPT-2 and GPT-3. Amodei has told the Senate of being one of the researchers who first documented the smooth, rapid improvement of AI systems as they scale, while working at OpenAI in 2018.47

Amodei co-founded Anthropic, which began at the start of 2021, and leads it as chief executive officer. Anthropic is a public benefit corporation whose purpose is the responsible development and maintenance of advanced AI for the long-term benefit of humanity, and Amodei serves on its board of directors.312

In 2023 Senate testimony, Amodei described the research Anthropic had conducted: constitutional AI, a method for training an AI system to behave according to a set of explicit principles; early work on red teaming, or adversarial testing of AI systems; and a series of foundational works in interpretability, the science of understanding why AI systems behave the way they do. Anthropic's constitutional AI paper, published in December 2022, explored enlisting AI systems to help supervise other AIs.78

Anthropic introduced Claude, a next-generation AI assistant, in March 2023, and in July 2023 launched Claude 2, which Amodei described as putting many of the company's safety innovations into practice as part of a "race to the top" on safety. In September 2023, Anthropic published its Responsible Scaling Policy, a series of technical and organizational protocols for managing the risks of increasingly capable AI systems.9710

In May 2024, Anthropic reported a significant advance in understanding the inner workings of AI models, identifying how millions of concepts are represented inside Claude Sonnet. In February 2025, Anthropic introduced Claude Code, a command-line tool for agentic coding.1112

Testifying before the Senate Judiciary Subcommittee on Privacy, Technology, and the Law on July 25, 2023, Amodei explained that if AI's risks can be mitigated, its benefits could be truly profound, from greatly accelerating treatments for diseases such as cancer to lowering the cost of energy and revolutionizing education.7

In the October 2024 essay "Machines of Loving Grace: How AI Could Transform the World for the Better", Amodei argued that most people are underestimating just how radical the upside of AI could be, envisioning a "country of geniuses in a datacenter" that could compress decades of progress in biology into a few years. "The Urgency of Interpretability" (April 2025) set out a goal of interpretability that can reliably detect most model problems by 2027, and "The Adolescence of Technology" (January 2026) argued that if humanity acts decisively and carefully, the risks of powerful AI can be overcome, with a hugely better world on the other side.131415