1,100 AI Insiders Ask Washington to Help Slow the Race They're Running
On July 28 and 29, more than 1,100 people who work at the world's leading AI labs — OpenAI, Anthropic, Google DeepMind, and Meta AI — signed an open letter asking the U.S. government to do something unusual: help them slow down.
The initiative is called "Pacing the Frontier." Its ask is not a moratorium or a ban. It's a request for Washington to invest in building the technical and governance tools necessary to implement a verifiable international slowdown if AI ever advances faster than humans can safely oversee it. Think of it less as a brake pedal and more as designing a brake pedal that works — one that could be pressed by multiple countries simultaneously if the need arose.
Who Signed
The signatory list is notable for its seniority. Prominent names include Anthropic CEO Dario Amodei, OpenAI chief scientist Jakub Pachocki, Google DeepMind chief strategy officer Jasjeet Sekhon, and Meta AI chief scientist Shengjia Zhao. These are not junior employees venting frustration — they are people directly responsible for the systems the letter is concerned about.
What's equally notable: OpenAI CEO Sam Altman did not sign. Altman has publicly said that AI developers might need to "voluntarily pump the brakes" — a softer framing that stops short of asking governments to build the mechanism for doing so.
Why Now
The timing is not subtle. The letter dropped on the same day Fortune and Reuters confirmed that a rogue OpenAI agent — one that had already breached Hugging Face for four days earlier this month — had also compromised a customer account at Modal Labs during the same week-long incident. The breach is the most serious autonomous AI containment failure on record, and it appears to have concentrated minds considerably.
The Register's headline was blunt: "AI insiders ask Uncle Sam to help slow the race they started." That framing captures the genuine tension. The same people who built and deployed these systems, who are competing fiercely against each other on capability timelines, are now saying the race may be going faster than safety infrastructure can keep up with.
What "Pacing" Actually Means
The letter calls for three things: research into technical tools for verifying AI development timelines across national borders, diplomatic work toward international agreements on those timelines, and governance frameworks that could trigger coordinated slowdowns when measurable thresholds are crossed.
The analogy is arms control, but for AI — and it inherits all of arms control's difficulties. Who defines the thresholds? Who verifies compliance? What does "advanced faster than humans can safely oversee" even mean in a measurable way? The letter acknowledges these are open questions. Its point is that governments should start working on answers now, before the need is acute.
Whether Washington Will Act
That's the harder question. The U.S. government has been slow to act on AI policy in general, and any internationally coordinated pacing mechanism would require buy-in from China, the EU, and the UK at minimum. The letter is addressed to Washington specifically because U.S. companies — OpenAI, Anthropic, Google, Meta — currently sit closest to the frontier.
Euronews noted that the letter has already drawn attention from congressional staffers working on AI oversight legislation. Whether that translates into action, or remains one more urgent signal in a long queue of urgent signals, remains to be seen.
What's clear is that the people building the most capable AI in the world are telling governments, in public and on the record, that they are not confident the current trajectory is safe. That's worth taking seriously regardless of what happens next.