Approximately 1,337 AI researchers and employees have signed a joint statement asking the United States government to support an international effort to “develop the technical and governance tools needed to deliberately pace the frontier of automated AI development.”
The statement includes senior representation from major Western AI labs: OpenAI, Anthropic, Google DeepMind, Thinking Machines, Meta and Safe Superintelligence Inc. Signatories reportedly include chief scientists and cofounders from Anthropic, Google and OpenAI, as well as the CEOs of Safe Superintelligence and Anthropic.
Content of the statement
The authors note that while AI could enable a substantially better future, that outcome is not guaranteed. They warn that leading AI companies believe they may be nearing a point where AI systems could automate aspects of AI research itself. Although it is difficult to predict precisely how much this would accelerate progress, the statement says there is a real risk that capability development could accelerate faster than humans can understand or control the resulting systems.
To realize AI’s benefits, the statement argues, industry, government and society may need the option to buy time to address emerging risks, develop security measures, and strengthen oversight. Yet each company and country faces intense competitive pressure not to unilaterally slow development, and today the world lacks the technical and governance tools to deliberately pace frontier-wide progress.
As a result, the signatories request that the U.S. government support an international initiative to create those technical and governance mechanisms to manage the pace of automated AI development at the frontier.
Why this matters
The statement frames the issue as a collective action problem: if AI becomes capable of improving itself or automating research (sometimes referred to as research self-improvement, RSI), the central challenges are social and institutional. Governments and companies must coordinate how the technology is developed and what mechanisms would allow society more time to adapt as systems increase in capability.
The authors suggest that some levels of machine intelligence could be, at least for now, too risky to pursue, and that public statements and international coordination are prerequisites for enabling society to address such risks responsibly.
Source and aims
The initiative is published under the title “Pacing the Frontier” on its official website, which hosts the full statement and list of signatories. By making the request to the U.S. government, the group seeks political backing for international work on technical monitoring, governance frameworks and other tools intended to manage the tempo of frontier AI development so potential harms can be better anticipated and mitigated.
