A debate over artificial intelligence safety at Salesforce’s Dreamforce conference highlighted a split between two of the sector’s most prominent executives. Anthropic CEO Dario Amodei and Nvidia CEO Jensen Huang appeared in San Francisco with Marc Benioff and laid out different views on how fast AI development should move.
Amodei spoke first during Dreamforce’s keynote, just days after publishing an essay calling on the industry to slow the pace of model development. He said that approach could help lead the sector forward by setting an example and showing that companies can hold themselves to a higher standard.
His proposal quickly drew support from parts of the tech industry. CNBC reported that OpenAI CEO Sam Altman, SpaceX CEO Elon Musk and Google DeepMind Chair Demis Hassabis all backed the idea publicly. Their comments followed warnings from researchers last week that AI could pose catastrophic risks. Amodei outlined a three-step plan aimed at tempering the rapid improvement of model capabilities without “sacrificing commercial advantage or the United States’ lead in AI.”
Huang offered a different perspective. Taking the stage with Benioff shortly after Amodei, the Nvidia chief said market forces are already in place and that the industry does not need new laws or regulations. In his view, speed and safety do not have to come at each other’s expense.
“It’s a false choice,” Huang said, arguing that companies can pursue both at the same time. He added that developers should move quickly but remain ready to pause if they believe a product is not safe or the situation is getting out of control. “Run as fast as you can,” he said. “But if you feel at any given point in time the company’s out of control or the product’s not going to be safe, take a pause and make sure you get it right.”
A day earlier, Huang was on stage at the All-In Summit in Los Angeles when he took a call from President Donald Trump. Speaking on speakerphone, Trump voiced concern about people trying to block the construction of AI data centers. Huang replied, “We’re not going to let that happen, sir.”
Dreamforce has been one of the software industry’s flagship events for more than a decade, drawing about 50,000 people into downtown San Francisco each year. Benioff’s keynote regularly features high-profile guests, and this year around 12,000 people were in the audience at Moscone Center.
The event also carried clear business implications. On Tuesday, Salesforce and Nvidia announced a new reasoning model for the Agentforce platform, built on one of Nvidia’s open-weight Nemotron models. Late last month, Salesforce and Anthropic expanded their strategic partnership and introduced “Claudeforce,” a new way for sales teams to access key data directly inside Anthropic’s Claude chatbot.
Benioff told reporters that the ultimate responsibility should rest with the model companies themselves. If they believe they need to slow down, he said, they should slow down; if they think they should move faster, they should do so, but they must then be held accountable for those decisions.
