This is Lex Fridman’s long-form interview with Dario Amodei, the co-founder and CEO of Anthropic, on Fridman’s podcast. Running over five hours, it is among the most detailed firsthand conversations available with the head of a frontier AI lab, and it doubles as a guided tour of how Anthropic thinks about building and governing powerful systems.
Amodei discusses the scaling laws that drove his early conviction that larger models would keep getting more capable, the safety concerns that led him to co-found Anthropic, and his views on alignment, interpretability, and the trajectory toward very capable AI. The episode also includes segments with Anthropic researchers Amanda Askell, who works on Claude’s character, and Chris Olah, who works on interpretability.
For a business or policy reader, this is a chance to hear an influential lab leader reason in detail rather than in soundbites. As the original Fridman interview on his own channel, it sits firmly in the first-person, primary-source category, and it pairs well with the more technical talks in this collection.