SAN FRANCISCO: Tech leaders offered sharply contrasting visions for the future of artificial intelligence safety Tuesday at Salesforce’s annual Dreamforce conference, exposing a growing rift between model builders advocating for a slower pace and executives championing market forces.
Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman aligned in calling for tempered development to address potential catastrophic risks, directly contrasting with Nvidia CEO Jensen Huang and Meta CEO Mark Zuckerberg, who argued against new regulations and slowing down.
The debate took center stage at San Francisco’s Moscone Center in front of about 12,000 attendees, following a weekend essay by Amodei urging the industry to slow model capability growth.
“I think that’s the way to lead the industry forward, to set an example, to say that everyone can always be better,” Amodei told Salesforce CEO Marc Benioff during the keynote. Amodei proposed a three-step plan to moderate development without sacrificing commercial competitiveness or U.S. leadership in AI.
Altman, appearing later in the day for a fireside chat with Benioff, endorsed Amodei’s stance, emphasizing that safety and monitoring must precede capabilities.
“I think it’s great for our industry to say we want to come together and we want to be able to coordinate and make sure we have enough time to do this safely,” Altman said, adding there should be "no qualifier" on a company's commitment to responsibility.
However, Huang dismissed the idea that speed and safety are mutually exclusive.
“It’s a false choice,” Huang said on stage. “You could definitely have both at the same time.”
Huang urged developers to "run as fast as you can," taking pauses only if a product feels out of control. In a subsequent CNBC interview, Huang called proposed AI regulations "completely unnecessary," pointing to existing product reliability laws.
Zuckerberg echoed Huang’s market-driven view in a Tuesday social media post, stating that fear of liability naturally incentivizes safety. He noted that Meta voluntarily delayed its Muse AI tools for safety reasons without government intervention.
"Trust and alignment are quickly becoming the most important capabilities that will differentiate agents and models," Zuckerberg wrote.
The public debate has also reached Washington. Speaking at the All-In Summit in Los Angeles a day earlier, Huang took a speakerphone call from President Donald Trump, who raised concerns about opposition to AI data center construction.
"We're not going to let that happen, sir," Huang told the president, who has previously criticized calls to slow AI progress.
Despite the philosophical split, business momentum continued. Salesforce and Nvidia announced a new reasoning model for Salesforce’s Agentforce platform built on Nvidia's Nemotron models, following a recent expanded partnership between Salesforce and Anthropic.
Addressing the divide, Benioff took a pragmatic stance: “If they feel like they should slow down, then they should slow down. If they feel they should speed up, they should speed up. And then they should be held accountable.”
