Employees from OpenAI, Anthropic, Google DeepMind, Meta, Microsoft, Mistral and other AI organizations have signed a public statement called Pacing the Frontier. The site lists 1,134 signatories and asks the US government to support an international effort to build tools that could deliberately pace automated AI development.
The list matters
The verified names include OpenAI chief scientist Jakub Pachocki, chief research officer Mark Chen, John Schulman, Wojciech Zaremba, Anthropic cofounder Jack Clark, Anthropic chief science officer Jared Kaplan, Claude Code lead Boris Cherny, Meta chief scientist Shengjia Zhao and Google DeepMind safety and alignment lead Anca Dragan.
The statement says leading AI companies may be close to automating AI research. Its request is narrower than an immediate pause: build the technical and governance mechanisms first, so society has an option to buy time if risks rise.
The hard question is enforcement
A voluntary slowdown could punish the first lab that complies. A government order could look like regulatory capture if it protects incumbent labs from open-weight or overseas rivals. That is why the useful policy questions are practical: who decides that a capability level needs a buffer, how compliance is audited, and whether companies in different countries are inside the same framework.
For readers outside the US, the China angle is direct. Kimi, GLM and Qwen have made open-weight competition a central part of the frontier-model debate. Any pacing regime that ignores open models, export controls and international access would be incomplete.
Sources: Pacing the Frontier public statement, The Verge, Business Insider, TechCrunch, CocoLoop; verified the 1,134-signatory count, key OpenAI/Anthropic/Meta/Google DeepMind signatories, automated AI research wording, request for US-backed international coordination, Hugging Face sandbox incident context and policy disputes over pacing.