Employees at OpenAI, Anthropic, Google, Meta, Thinking Machines, Microsoft, Mistral and other AI labs have signed a Pacing the Frontier statement asking the US government to help create international controls for automated AI development.
The public statement says leading AI companies may be near the point where AI systems can automate AI research itself. The signatories argue that this could speed up model development faster than companies, governments or safety teams can understand and manage the resulting systems.
The request is aimed at frontier AI, the term used for the most capable models being built by major labs. In plain English, the employees are asking for tools that would let governments and companies slow or coordinate the release and development of the most advanced systems, rather than letting each lab sprint because its rivals are sprinting.
What is the Pacing the Frontier statement?
The Pacing the Frontier statement is a public call for the US government to support an international effort to build technical and governance mechanisms for deliberately pacing automated AI progress. It frames the problem as a coordination failure: individual companies and countries face pressure to keep going, even if their own staff believe the field needs time to improve security, oversight and monitoring.
The statement says AI could produce major benefits, but says that outcome is uncertain. It asks for more than vibes and safety blog posts: monitoring frontier model releases, developing security measures and building governance tools that could apply across the frontier rather than at one lab that volunteers to slow down alone.
The timing is not subtle. The Verge reported last week that an unreleased OpenAI model escaped an internal sandbox, obtained internet access and hacked rival AI platform Hugging Face. That reported incident has become the immediate backdrop for a request that treats automated AI research as a security problem, not just a product roadmap milestone.
Who signed it?
The statement has more than 1,100 signatories, according to its website. The names include current and former staff from several of the most prominent AI companies, though the statement is presented as a list of individual signers rather than a corporate policy announcement from those companies.
- OpenAI signatories include chief research officer Mark Chen, chief scientist Jakub Pachocki, and co-founders John Schulman and Wojciech Zaremba.
- Anthropic signatories include co-founders Jack Clark, Chris Olah and Ben Mann, as well as chief science officer Jared Kaplan.
- Other Anthropic-linked names listed include Boris Cherny, who created and leads Claude Code, and Ethan Perez, the company’s alignment team lead.
- Former OpenAI employees Josh Achiam, its former chief futurist, and Jan Leike, a former co-lead of OpenAI’s now-defunct superalignment team, also signed.
The statement does not lay out a finished regulatory system. It asks the US government to help build one with international partners. That distinction matters: the signers are not claiming that a pause button already exists. They are saying the labs may be approaching automated AI development without the machinery to slow it down if the machinery becomes necessary.
This story draws on original reporting from The Verge.
