Dario Amodei

Dario Amodei (born 1983) is an American AI researcher and entrepreneur who cofounded and serves as chief executive of Anthropic. Trained as a physicist and computational neuroscientist, he led the development of GPT-2 and GPT-3 as vice president of research at OpenAI, then left in late 2020 with his sister Daniela Amodei and a group of OpenAI colleagues to found a competing lab built around the thesis that advanced AI should be developed under strict safety constraints. He emerged from the Effective Altruism milieu, took the Giving What We Can pledge, and is the author of the October 2024 essay "Machines of Loving Grace."¹²

Education and Early Career

Amodei was born in San Francisco to Riccardo Amodei, an Italian-American leather craftsman from Tuscany, and Elena Engel, a Jewish-American library project manager from Chicago. He attended Lowell High School, began his undergraduate study in physics at Caltech, and transferred to Stanford University, where he took a bachelor's degree in physics in 2006. He completed a PhD in biophysics at Princeton University in 2011 as a Hertz Foundation Fellow, with a thesis titled "Network-Scale Electrophysiology: Measuring and Understanding the Collective Behavior of Neural Circuits" supervised by Michael Berry and William Bialek, and won the Hertz thesis prize. He then did postdoctoral work at the Stanford University School of Medicine.¹³

His move into machine learning ran through industry research labs. He worked at the Chinese search company Baidu from November 2014 to October 2015 under Andrew Ng, where his work on the relationship between model performance, data, and compute contributed to the empirical "scaling laws" that later became central to his worldview, and then at Google Brain as a senior research scientist. In 2016 he joined OpenAI, where over the next four years he rose to vice president of research and helped pioneer reinforcement learning from human feedback, the alignment technique that underpins modern conversational models.¹³

OpenAI and the Departure

At OpenAI, Amodei directed the team that built the GPT-2 and GPT-3 language models, the latter consuming a reported 50 to 60 percent of the company's compute and tens of millions of dollars to train. He also ran the safety team and co-authored the 2020 "Scaling Laws for Neural Language Models" paper, which formalized the observation that model capability grows predictably with size, data, and compute. That empirical result became the foundation of his conviction that simply scaling up models would keep producing more powerful systems, a thesis he carried into Anthropic.³⁴

A faction around Amodei came into conflict with OpenAI's leadership over the direction of the company, the pace of commercialization, and the 2019 Microsoft investment and capped-profit restructuring. One colleague later described the period as one in which roughly half the team's time went to arguing internal disagreements rather than research. In December 2020 Amodei resigned, and over the following months he and a group of senior OpenAI staff, including his sister Daniela, departed to start their own lab; the company has been described as founded by seven former OpenAI employees, with the broader founding cohort including Jared Kaplan, Jack Clark, Chris Olah, Tom Brown, Sam McCandlish, and Ben Mann.¹⁴

Anthropic and the Scaling Thesis

Amodei incorporated Anthropic in early 2021 as a public-benefit corporation, with Daniela as president, and the two raised money from funders tied to effective altruism, including Open Philanthropy, whose cofounder Holden Karnofsky is married to Daniela. In 2022 the company took a roughly 500-million-dollar investment from Sam Bankman-Fried through Alameda Research months before the FTX collapse, and later raised tens of billions from Amazon and Google. Amodei has positioned Anthropic publicly as the safety-conscious frontier lab, championing a "Responsible Scaling Policy" of capability thresholds even as the company competes directly at the frontier.²⁵

Amodei is a prominent voice on both AI risk and AI optimism. In October 2024 he published "Machines of Loving Grace," a roughly 14,000-word essay arguing that "powerful AI," which he likens to "a country of geniuses in a datacenter," could compress decades of progress in biology, neuroscience, economic development, governance, and the meaning of work into five to ten years; he opened it by stating that "most people are underestimating just how radical the upside of AI could be, just as I think most people are underestimating how bad the risks could be." He has separately warned that AI could eliminate a large share of entry-level white-collar jobs, testified and written on the danger of authoritarian misuse, and argued for export controls to keep advanced chips from China. Forbes estimated his net worth at roughly 15.5 billion dollars in mid-2026 as Anthropic's valuation climbed.⁶⁷⁸