More than a thousand OpenAI, Google and Meta staff signed a letter asking for a brake on AI
The Pacing the Frontier letter, signed by over 1,100 AI lab employees, asks Washington to build the tools for a coordinated, verifiable slowdown in AI development.
The people building the machines are asking for a brake. An open letter that began circulating on 28 July has gathered more than 1,100 signatures from employees at OpenAI, Anthropic, Google and Meta, and it asks the US government to help assemble what they call a pacing mechanism for artificial intelligence.
What does the Pacing the Frontier letter actually ask for?
Not for anything to be switched off tomorrow. It asks for infrastructure: the technical tools and the governance rules that would make it possible, if it ever becomes necessary, to slow development down in a coordinated way across countries and — crucially — in a verifiable way, meaning there is a means of confirming that whoever promised to slow down actually did. That is the difference between an agreement and an agreement you can check.
Is this a pause on artificial intelligence?
No, and that is the part that keeps getting lost. The signatories, who include chief scientists at the major labs, say the risk on their minds is recursive self-improvement: the point at which a system can develop itself faster than people can follow what is happening. They want the brake built before it is needed, not halfway down the hill.
What gives the letter its weight is who lined up behind it. Anthropic’s Dario Amodei signed; Sam Altman has been more reserved, but both companies formally backed the initiative. Watching direct rivals jointly ask someone to impose a limit on them is not an everyday sight in this industry — and it lands days after a containment failure at one lab made the argument a lot less theoretical.
Europe is already on a different track: instead of a voluntary mechanism, there are written rules, with transparency and risk-assessment duties set out in the EU’s artificial intelligence regulation. The global argument has been heating up on every side, as it did when Xi Jinping opened China’s biggest AI conference with a pointed message about cooperation.
Then comes the hard part: a letter is not a law, and Washington has said nothing concrete yet. But when the engineers themselves start asking for a brake pedal, it is worth asking what they can see from the front seat.
By Oliver Grant
Image: HaeB / Wikimedia Commons (CC BY-SA 4.0)
.jpg?width=1280)
