Amodei and Huang Split on AI Safety Pace at Dreamforce

Anthropic CEO Dario Amodei and Nvidia CEO Jensen Huang offered contrasting views on how fast artificial intelligence should advance, appearing back-to-back with Salesforce CEO Marc Benioff at the Dreamforce conference in San Francisco.

Amodei spoke Tuesday before roughly 12,000 people at the Moscone Center, days after publishing an essay urging the industry to slow model development. He told Benioff that leading by example matters, saying everyone can always be better. His proposal followed warnings from industry researchers about AI's potential for catastrophic harm, and he outlined a three-step plan to temper how quickly model capabilities improve without sacrificing commercial advantage or the United States' lead in AI.

Huang, whose company is the world's most valuable, took the stage shortly after and rejected the idea that speed and safety must be traded off. Market forces already provide discipline, he said, and no new laws or regulations are needed. He called the tension a false choice, adding that developers should run as fast as they can but pause if a product isn't safe.

The day before, Huang had appeared at the All-In Summit in Los Angeles, where he took a speakerphone call from President Donald Trump, who voiced concern about people trying to block AI data center construction. Huang replied, "We're not going to let that happen, sir."

Benioff, asked by reporters about the pace question, said companies should follow their own judgment and then be held accountable. OpenAI CEO Sam Altman, speaking with Benioff at the conference, said safety and monitoring must come before capabilities and that responsibility should carry no qualifier.

Salesforce and Nvidia also announced a reasoning model for Salesforce's Agentforce platform, built on an Nvidia open-weight Nemotron model. Salesforce and Anthropic recently expanded their partnership with a tool called Claudeforce.

Source: CNBC