Anthropic CEO Says “Slow Down AI”

In this episode, we examine a striking proposal from Dario Amadei, CEO of Anthropic, one of the leading companies building frontier AI systems. Amadei has just made one of the strongest public arguments for slowing down the pace of AI development. What makes this particularly noteworthy is that Amadei is not anti-AI. He believes artificial intelligence could cure major diseases, accelerate economic growth, and dramatically improve human life. Yet he warns that the risks are becoming more serious, including loss of control, cyber attacks, bioterrorism, and major economic disruption. His central concern is recursive self-improvement: AI systems increasingly helping to build the next generation of AI systems. He argues this could accelerate capability growth beyond our ability to understand or control it. Amadei's proposal has three key parts: embedding independent evaluators inside AI companies with employee-like access to verify safety practices, coordinating pacing among democratic countries, and eventually achieving global coordination. He emphasizes that pacing should be tied to capabilities rather than arbitrary timelines, with checkpoints where more powerful systems must meet higher safety standards before progressing further. Join us as we break down why one of AI's leading figures believes the frontier needs to be paced, what he is proposing, and whether slowing down is actually realistic in a world where companies and countries compete for AI supremacy.