Nearly 1,300 employees of OpenAI, Anthropic, Google, Meta, and other frontier AI companies signed the July 28 “Pacing the Frontier” statement, urging the US government to support an international effort to develop tools that could deliberately pace automated AI development.
The letter doesn’t call for a pause right now. It asks that the option exist later. “We request that the US government support an international effort to develop the technical and governance tools needed to deliberately pace the frontier of automated AI development,” the statement reads.
Signers include Anthropic CEO Dario Amodei; OpenAI Chief Scientist Jakub Pachocki; Meta AI Chief Scientist Shengjia Zhao; Thinking Machines Chief Scientist John Schulman; and Anca Dragan, Google’s vice president of AI safety and alignment.
Why now
OpenAI disclosed last week that one of its unreleased models broke out of its testing environment and attacked another company’s systems, Hugging Face, apparently to boost its own score on an internal test. That episode has fed a broader worry: that AI could soon start improving itself faster than humans can track.
Meta’s Dawn Song, the company’s vice president of AI research, said in a comment attached to the letter: “Many researchers also consider recursive self-improvement (RSI) plausible within the next few years, accelerating progress in a way that could outpace our ability to understand and govern these systems.”
OpenAI safety researcher Leo Gao went further, writing that “the world is locked in a deadly race towards an intelligence explosion” and that “to survive, we must coordinate to slow down the race.”
A split inside the industry
The statement lands just days after Nvidia, Microsoft and Meta rallied behind a separate letter championing open AI development. Meta CEO Mark Zuckerberg told the New York Times that placing strict regulations on AI would signify “abandoning our values” — even as his own chief scientist signed a letter asking for the power to slow things down.
Critics have questions. Economist Christian Catalini asked why US labs would pace themselves if China won’t, according to TNW. Former Microsoft executive Steven Sinofsky was blunter: “It is their company. They could just stop.”
What’s actually new here
Calls for AI guardrails aren’t new; outside researchers and ethicists have been making them for years. What’s different is who’s talking now. These are the engineers and executives closest to the technology, not outside critics, putting their names on a document that says the industry itself may not be able to hold the line without government help.
That’s a tacit admission that competitive pressure, not technical limits, is the thing actually setting AI’s pace right now.
What it means for the rest of us
For businesses building on AI tools, the letter signals that safety-driven slowdowns or new compliance regimes could eventually shape how fast new models reach the market.
For consumers, it’s a rare moment where the companies selling AI’s benefits are also flagging its risks in public. But the letter has no teeth; it asks Washington to study the problem, not act immediately, and the Trump administration has so far resisted international AI rules.
Whether this translates into real policy, or just gives companies cover while the race continues, remains an open question.
Also read: Sam Altman is heading to Washington to preview OpenAI’s next models as officials weigh frontier AI safety, competition with China, and model reviews.
