More than 1,100 employees sign an unusual open letter

On July 28, 2026, more than 1,100 current employees at leading AI companies including OpenAI, Anthropic, Google DeepMind, and Meta reportedly published a joint open letter titled "Pacing the Frontier." The letter asks the US government to support an international effort to build technical and governance tools that could deliberately slow the pace of frontier AI development, if that ever becomes necessary. It is a notably large and cross-company show of concern from people who work inside the very labs racing to build increasingly capable AI systems, and it arrives at a moment when "frontier AI" — the most advanced, most capable models being trained today — has become a regular subject of public debate over safety and control.

Among the reported signatories are Anthropic CEO Dario Amodei and several of the company's co-founders, OpenAI chief scientist Jakub Pachocki, OpenAI chief research officer Mark Chen, Meta AI chief scientist Shengjia Zhao, and Google's VP of AI safety Anca Dragan. The presence of senior researchers and executives from rival labs on the same document is being described as notable in itself, since these companies are otherwise fierce competitors racing to ship more capable models before one another, often within weeks of each other's releases.

Not a call to pause now, but a request to prepare

Crucially, the letter does not ask any company to slow down or pause AI development today. Its central line, according to multiple reports, reads roughly: "We request that the U.S. government support an international effort to develop the technical and governance tools needed to deliberately pace the frontier of automated AI development." The stated goal is to have mechanisms ready in case AI capabilities ever advance faster than humans can safely oversee them — a kind of insurance policy rather than an emergency brake pulled today. Reports describe this as an attempt to separate the question of "should we build the tools to slow down" from the much more contested question of "should we slow down right now," which the signatories are reportedly not trying to settle in this letter.

Within hours of the letter's publication, both OpenAI and Anthropic reportedly endorsed it at the company level, a rare instance of two fiercely competing labs backing the same public statement on AI safety policy. Anthropic's research on recursive self-improvement, published the previous month, had already pointed to the need for tools that let society deliberately pace AI progress, and some observers see the letter as a natural extension of that thinking into a concrete, public policy ask directed at Washington rather than at any single company's internal roadmap.

Signed against a backdrop of rising industry tension

The letter emerged during a period of heightened concern inside the AI industry over safety and governance, reportedly including unexpected behavior inside a rival lab's internal systems and broader worries about the security risks posed by increasingly autonomous AI systems. Against that backdrop, employees choosing to publicly ask their own government to prepare a slowdown mechanism — rather than waiting for company leadership to act on its own — is being read by some commentators as a shift in how AI workers see their own responsibility for the technology they are building day to day, and as a sign that internal safety conversations at these labs are increasingly spilling out into public view.

Reported signature counts have varied somewhat by outlet, ranging from just over 1,100 to as many as 1,200 or more as the letter continued to circulate and gather names after publication. The figure most consistently confirmed across outlets is "more than 1,100," and the total is reportedly still climbing as more employees, including some outside the four companies named above, add their names to the list.

What happens next

The letter carries no legal weight and does not obligate any government or company to act. Whether Washington moves toward building the kind of international coordination framework the signatories describe remains to be seen, and no formal government response had been reported as of publication. What is clear is that a large group of engineers and researchers working at the very companies racing to build more capable AI systems chose to ask, in public, for a plan to slow that race down if it ever outpaces human oversight — without asking anyone to slow down yet, and without waiting for a crisis to make the case for having a plan in the first place.