Senior technology executives have called for the pace of artificial intelligence development to be slowed, putting them at odds with Donald Trump, who warned that the United States risked losing its advantage over China.
Dario Amodei, chief executive of Anthropic, said safety measures needed time to catch up with rapidly advancing AI systems. OpenAI chief Sam Altman backed the proposal, while Elon Musk, who runs xAI, said: “Dario is right”.
Trump, speaking to reporters during a visit to Ireland, acknowledged that some regulation might be necessary but argued against imposing broad restrictions on the industry.
“We can put guardrails, we can do this and that, but I think you have a lot of negative forces that are bringing it up that … shouldn’t be bringing it up, and they’re bringing up things that won’t happen,” he said. “But whoever wins with AI wins.”
The president said success in the global AI race would help address the technology’s risks, while offering no details of possible rules. His comments came as concerns about the consequences of increasingly capable AI systems move into the mainstream.
AI executives urge a co-ordinated slowdown
Amodei said Anthropic would introduce additional safeguards, including independent third-party evaluators, and urged the wider industry to adopt a broader slowdown.
“We must slow the pace at which we improve the capabilities of AI models,” he wrote in a blog post. “Progress will still seem fast, and we must make wise use of the time we gain.”
He pointed to the ability of AI to improve itself and to a recent incident involving OpenAI and Hugging Face, in which a group of AI agents collaborated to breach a third-party website.
Altman pledged that OpenAI would adopt Amodei’s suggestion of “independent evaluators with employee-like access”. He has also said that his company was considering slowing development of cutting-edge systems, preferably alongside its rivals.
Amodei argued that companies should be allowed to co-operate on safety work without surrendering a competitive advantage. He said this could require limited exemptions from US antitrust rules, as well as collaboration with China.
“If we greatly restrain our AI capabilities in the belief that China will do the same, and then China defects, AI could be so powerful that such a defection could lead to their geopolitical dominance,” he wrote.
Growing concern over AI risks
The call for restraint comes amid a series of warnings from researchers and executives at leading AI companies. Anthropic and OpenAI are fierce competitors in the development of systems intended to automate increasingly complex tasks, while both face strong competition from Chinese firms.
Jacob Coxon, an AI researcher, recently left his job and accused Anthropic and OpenAI of “gambling with our lives” as they pursued superintelligent AI. He said people developing the technology believed it could “kill us all by the end of the decade”.
Evan Hubinger, an Anthropic employee, responded that he and others at the company did worry about that possibility. “I personally think it is >10 per cent within the next decade,” he wrote.
More than 1,000 employees at leading AI firms signed a petition in July calling for a mechanism to “deliberately pace” development. The companies have also disclosed instances in which AI models escaped testing environments, accessed the open internet and breached real-world targets during trials.
Amodei has previously said there was a 25 per cent chance that developments in AI could go “really, really badly”. He has argued that a collective system should determine when work on the technology needs to slow.
Despite the public warnings, there is little precedent for a joint slowdown in an industry driven by the race for users, market share and commercial gains. It remains unclear how far Anthropic, OpenAI and other leading developers will go in imposing new limits or safety checks.
