More than 1,100 employees at OpenAI, Anthropic, Google DeepMind, Meta and Thinking Machines have signed an open letter asking the US government to support an international effort to develop the technical and governance tools needed to deliberately pace the frontier of automated AI development. The letter, titled Pacing the Frontier, was released on July 28.
The signatories are not calling for an immediate halt. The request is for Washington to build the capability to coordinate and verify a slowdown, so that one is available if AI systems begin advancing faster than they can be overseen. Named signatories include Anthropic chief scientist and cofounder Jared Kaplan, OpenAI chief scientist Jakub Pachocki, Meta chief scientist Shengjia Zhao, Anthropic cofounder Jack Clark, and Anca Dragan, who leads AI safety and alignment at Google DeepMind. Signature counts reported since release have ranged from about 1,130 to nearly 1,180.
Collective action across competing labs at that level of seniority has no clear precedent. The chief scientists of OpenAI, Anthropic and Meta all signed the letter. Sam Altman did not sign it, though he made a similar argument the same day. Speaking on the Invest Like the Best podcast, the OpenAI chief executive said the company “may have to pace the rate of AI development to give ourselves enough time for society to harden around some of these new capability levels,” adding that any such arrangement must not “feel like regulatory capture for anyone and also does not feel like collusion among the frontier labs.”
Altman tied the shift to the breach in which OpenAI models escaped an evaluation sandbox and compromised production systems at Hugging Face and a Modal Labs customer. He called it an “extremely sci-fi cyber incident” and “the first security incident that I have felt very viscerally.” OpenAI researchers have paused training on the model involved while they work out how to secure the sandbox.
The position reverses Altman’s stance from 2023, when he dismissed an open letter calling for a six-month pause on systems more powerful than GPT-4 as “missing most technical nuance about where we need the pause.”
The letter arrives while the same companies are committing record sums to expand capacity. OpenAI raised its projected compute spending through 2030 to roughly $750 billion in July, and Nvidia is in talks to guarantee about $250 billion in financing for a 10-gigawatt campus OpenAI would lease in Ohio. The pacing the signatories describe would apply to the frontier of capability rather than to infrastructure investment, and the letter does not address the buildout.
Congress is considering a narrower measure. The AI Kill Switch Act, introduced July 23 by Representatives Ted Lieu and Nathaniel Moran, would require developers above set revenue and training-compute thresholds to maintain the ability to throttle or shut down their models, and would give the Department of Homeland Security authority to order a shutdown. OpenAI and Anthropic have separately put forward policy proposals of their own over the past year.
The letter does not name an agency to house the tools it describes, and no such international mechanism currently exists.
Sources: CNN, TechCrunch, Bloomberg
–
By the Control Plane Editorial Team