On Tuesday morning in San Francisco, two of the most powerful figures in artificial intelligence shared the same conference stage and reached opposite conclusions. Anthropic's Dario Amodei, whose company builds the Claude family of AI models, renewed his argument that the industry must decelerate and submit to external scrutiny. Nvidia's Jensen Huang, whose chips power virtually every major AI system on the planet, offered a two-word rebuttal: "Run as fast as you can." Amodei framed his case through an automotive analogy.

When one car manufacturer discovers a brake failure, he argued, the responsible response from every other manufacturer is not to attack the rival but to audit its own safety record. He outlined three concrete steps he believes the industry should take: embedding independent, third-party evaluators inside AI laboratories; coordinating safety standards among democratic nations; and, eventually, building a broader international framework. Anthropic, he said, has already committed to the first of those measures and intends to open a dialogue with competitors on the remaining two. His position echoes recent statements from Elon Musk of xAI and OpenAI's Sam Altman, and follows the resignation of former Anthropic researcher Jacob Coxon, who warned publicly that AI could threaten human extinction within a decade.

Huang rejected almost every premise of that argument. He told the Dreamforce audience that existing regulatory intervention is unnecessary, suggesting instead that companies should simply refrain from releasing products until those products are demonstrably safe. He dismissed fears about AI-driven unemployment as "completely nonsense" and predicted that every enterprise on earth would eventually redefine itself as an AI company.