For people using artificial intelligence agents, the dispute turns on whether companies can be trusted to make systems safe while pushing ahead with development. Mark Zuckerberg, Meta’s chief executive, and Jensen Huang, Nvidia’s chief executive, say safety can be driven by competition, legal liability and company decisions. Dario Amodei, the chief executive of Anthropic, has called for AI companies to deliberately slow the pace of capability development.
Zuckerberg set out his position in a social media post on September 15, 2026. He wrote: “There is a lot of debate about slowing progress on capabilities until alignment catches up.” He has rejected calls to slow the development of sophisticated artificial intelligence models, arguing that companies have reasons of their own to ensure their products behave in ways users want.
“people won't want to use agents that are misaligned with them,” Zuckerberg said. “labs have a strong natural incentive to make their models more aligned.” He said trust and alignment, a term used for ensuring AI systems act consistently with intended goals, would increasingly distinguish competing products.
“My view is that trust and alignment are quickly becoming the most important capabilities that will differentiate agents and models. Any lab that doesn't focus on alignment will fall behind,” Zuckerberg said.
He said companies should decide for themselves how to manage safety during development. “Every lab has the responsibility and incentive to move at the pace required to train its models safely, and the ability to take its own actions to ensure that happens,” he said. Zuckerberg also said “labs face significant liability if their models cause harm, so they have a strong incentive to prevent this as well.”
As an example of voluntary restraint, Zuckerberg said Meta held back its personal AI agent, Muse, “for several months to focus on safety and security.” He said Meta “didn't call for everyone else to do this before we would” and had “just did it as part of our day-to-day work because it was clearly the right thing for people and for us.” Zuckerberg did not give details of the safety and security work carried out during the delay.
Zuckerberg’s argument places him close to Huang, who also rejects the idea that safety requires a general reduction in development speed. Speaking on Tuesday at the annual Salesforce Dreamforce event, Huang described the choice between rapid progress and safe products as “a false choice.”
In a conversation with Salesforce chief executive Marc Benioff, Huang said: “Run as fast as you can. 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.”
Huang also opposed calls for greater government involvement. On CNBC’s “Mad Money” on Tuesday, he called the push for more government intervention “just completely unnecessary” and said there were already “plenty of laws.” Zuckerberg and Huang have both dismissed recent calls for government regulations, saying companies should retain responsibility for the issue.
The disagreement centres on whether competition can supply sufficient discipline in an industry whose products are advancing quickly. Zuckerberg’s view is that alignment can create an advantage in the market because users will favour systems they trust. Huang similarly presents safety as compatible with speed, provided companies pause where they judge that a product is unsafe or development has become difficult to control.
Amodei takes the opposing approach. He has argued that companies should deliberately slow capability development so that safety research can keep pace. The brief does not state whether he or other advocates of a slower approach have responded directly to Zuckerberg’s and Huang’s remarks.
Zuckerberg has also backed outside scrutiny as part of companies’ safety work. He called third-party testing and auditing of models “industry best practice” and said “it would be helpful for there to be a larger and more diverse ecosystem of evaluators.”
Trump has also criticised those who favour pacing AI development. “The robots will not be taking over. The AI will not be taking over the rest of the world. The whole thing is a hoax,” he said. He said advocates of a slower pace were “just playing right into the hands of a lot of people that don't want to see it happen. That could be political people, that could also be China.”
Huang told Trump: “we're going to make sure that everybody wins in the AI race of America. Every industry, every company, every state, every people.”
The competing positions leave a central question unresolved: whether companies’ commercial and legal incentives, together with voluntary testing, can keep safety work in step with AI capability development, or whether the industry should collectively slow that development.
Comments (0)