Trump Rejects the AI Slowdown Call: Whoever Wins With AI, Wins
President Trump has rejected the slowdown call from the CEOs of Anthropic, OpenAI, and xAI, saying whoever wins with AI, wins and insisting the technology does more good than bad. His response came as global tech stocks slid for a second session, after the industry's own leaders jointly asked for permission to slow frontier development down.
On this page
The answer from Washington is no
The question the AI industry asked its government this weekend, may we please slow down, has received its answer. President Trump rejected the slowdown call on Sunday, saying "whoever wins with AI, wins," and insisting the technology does "more good than bad." The response closes, for now, any expectation that the United States would institutionalize the pacing framework its own leading chief executives asked for.
The rebuff lands on the second day of a story that has moved unusually fast. On Saturday, Anthropic CEO Dario Amodei published an essay arguing the industry must slow the pace at which it improves AI models, and by Sunday his rivals Sam Altman and Elon Musk had both publicly backed the call, Musk agreeing in one line.
Markets kept sliding while the White House said no
The selling continued into Monday. Global AI-related stocks slid for a second session, with US markets opening lower and technology shares leading the decline, as investors priced in what a genuine slowdown would do to the spending plans behind the sector. The BBC carried live coverage of the declines through the day, and CNBC framed the slide as the market's reaction to the united front from Amodei, Altman, and Musk.
The irony is structural. The CEOs asked for permission to slow down precisely because competitive pressure will not let any of them slow down alone. The President's answer preserves that competitive pressure, and with it the dynamic Amodei warned about: each lab keeps accelerating because standing still means losing.
What the rejection means for safety commitments
It would be wrong to say nothing changes. Anthropic and OpenAI have both pledged to admit outside evaluators, Musk has floated cross-lab model reviews before launch, and a Senate subcommittee has OpenAI on a document deadline for October 1. Those commitments exist independently of the White House's enthusiasm, and the summer's incidents, from the Hugging Face breach to the agent back-channels, gave them an evidence base no press release could.
But the federal lever is officially off the table. The administration's position, maximum speed and minimal friction, means voluntary lab commitments, market reactions, and the mid-September dialogue with China are the only remaining mechanisms. The Beijing talks, expected this week with AI-directed cyberattacks on the agenda, will now proceed with Washington committed to outpacing, not restraining, the technology under discussion.
The practical takeaway for readers
For people deciding how much to trust AI tools, the signal from this weekend is that no external brake is coming from Washington. Safety pacing, where it happens, will be self-imposed by labs with mixed incentives, and enforcement of failures will be reactive rather than preventive. That argues for the habits this site keeps coming back to: know what your AI tools send and store, keep sensitive work on local models where the choice is yours, and treat vendor safety claims as claims to verify rather than guarantees.
The slowdown debate will continue, with Amodei's essay as its founding document and the midterm elections looming in November. But the race, as the President framed it, is still on.