At a San Francisco technology conference on Tuesday, Anthropic's chief executive, Dario Amodei, renewed his call for AI development to slow down, while Nvidia's chief executive, Jensen Huang, argued the industry should accelerate. Speaking at Salesforce's Dreamforce event, Amodei used a car-safety analogy to urge firms to scrutinise their own practices rather than attack rivals over incidents, while Huang countered that companies should simply "run as fast as you can" and avoid new regulation. The exchange highlights a deepening divide among Silicon Valley leaders over how quickly artificial intelligence should be developed and deployed.

Amodei proposed three measures: embedding independent evaluators within AI firms, coordinating safety standards among democratic nations, and eventually pursuing broader global coordination, saying Anthropic had already committed to outside evaluators. Huang, who also addressed the conference, dismissed fears that AI would destroy jobs and said Donald Trump had called him at a separate event the previous day to describe AI concerns as a "hoax". OpenAI's Sam Altman, speaking later, called for greater rigour on security following a hacking incident involving AI agents, saying the public was "right to be afraid" of the technology's risks, though he distanced himself from Amodei's slowdown framing. The debate follows warnings from a former Anthropic researcher, Jacob Coxon, who resigned warning AI could risk human extinction within a decade.