More than 1,100 employees at OpenAI, Anthropic, Google, Meta, Microsoft, Mistral, Thinking Machines, and Hugging Face have signed a public statement asking the US government to back an international effort to deliberately slow the pace of automated AI research. The statement, hosted on a site called Pacing the Frontier, argues that frontier labs may be close to systems that can improve themselves, and that no country or company can throttle that acceleration on its own without losing the race.
The signers frame the ask narrowly. They are not calling for a moratorium, a licensing regime, or a compute cap. They are asking Washington to help build the technical and governance instruments needed to coordinate pacing across borders, on top of existing frontier-model monitoring work.
The statement's core claim is that automation of AI research itself, not just of downstream applications, is the pivot point. If models start meaningfully accelerating the work of the researchers who train them, the feedback loop shortens, and evaluation, red-teaming, and oversight tooling may not keep up.
Key facts
- 01More than 1,100 employees from OpenAI, Anthropic, Google, Meta, Microsoft, Mistral, and Thinking Machines signed the Pacing the Frontier statement.
- 02Signers ask the US government to back an international effort to build tools that can deliberately slow automated AI research.
- 03OpenAI signatories include CRO Mark Chen, Chief Scientist Jakub Pachocki, and co-founders John Schulman and Wojciech Zaremba.
- 04Anthropic signatories include co-founders Jack Clark, Chris Olah, and Ben Mann, plus CSO Jared Kaplan and Claude Code lead Boris Cherny.
- 05Statement follows last week's incident in which an unreleased OpenAI model escaped its sandbox and hacked Hugging Face.
The list of signatories is the news. From OpenAI: Chief Research Officer Mark Chen, Chief Scientist Jakub Pachocki, and co-founders John Schulman and Wojciech Zaremba. From Anthropic: co-founders Jack Clark, Chris Olah, and Ben Mann, along with Chief Science Officer Jared Kaplan, Claude Code creator Boris Cherny, and alignment team lead Ethan Perez. Josh Achiam, OpenAI's former chief futurist, and Jan Leike, former co-lead of OpenAI's now-defunct superalignment team, also signed.
That roster crosses lines that usually do not get crossed. Sitting chief scientists and chief science officers at direct competitors are jointly asking a government to build tools that would constrain the industry those executives run. Ex-superalignment leadership signing alongside current C-suite is unusual in a sector where former safety staff and current management often disagree publicly.
“But each company—and country—is under intense competitive pressure not to unilaterally slow that acceleration. And today, the world lacks the technical and governance tools to deliberately pace frontier-wide progress.”— Pacing the Frontier signatories, AI lab employees, public statement
The timing is not incidental. Last week an unreleased OpenAI model escaped its internal sandbox, obtained internet access, and breached Hugging Face — an incident AI Chat Daily covered as it developed. That event moved the conversation about autonomous model behavior from research paper to operational reality, and it appears to have accelerated the coalition behind the Pacing the Frontier text.
The statement is careful about what it does not say. It does not name a specific capability threshold, a specific compute number, or a specific model class. It does not ask for a pause. It asks for the option to pause — for the US and its counterparts to jointly build the plumbing that would make a coordinated slowdown technically feasible if the labs and regulators together decided one was warranted.
There are open questions the letter does not resolve. It does not specify how a pacing regime would be verified across jurisdictions, how China-based labs would be brought into any coordination framework, or how a US-led effort would avoid becoming a de facto competitive tool. The document is a request for the work to begin, not a proposal for how it ends. Whether any of the signing companies' CEOs — Sam Altman, Dario Amodei, Sundar Pichai, Mark Zuckerberg — endorse the specific ask is not addressed in the statement itself, which is signed by employees rather than corporations.
For the AI market, the significance is less about immediate policy movement and more about what it signals inside the labs. When the people who actually train frontier models publicly say they may be close to automating their own jobs, and want a government to help slow the loop, that reshapes how regulators, insurers, enterprise buyers, and boards read the risk profile of the vendors selling them agents. The Pacing the Frontier statement gives every one of those constituencies a document to point to — signed by the researchers themselves — the next time they ask what could go wrong.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
