Amodei's Departure from OpenAI

Amodei's departure from OpenAI in the spring of 2021 was the founding event of Anthropic and a pivotal moment in the history of AI safety as an institutional commitment. Amodei had joined OpenAI in 2016 and risen to vice president of research, overseeing some of the most consequential capability advances in the field — including the scaling experiments demonstrating that larger models trained on more data exhibited qualitatively new capabilities. His departure took with him not only his own expertise but a cohort of senior researchers who shared his concerns. The reason, as Amodei later described it, was the widening gap between what frontier AI organizations said about safety and what they did about it — a structural problem rooted in incentive systems rather than individual malice.