A joint statement signed by 1,132 employees at frontier AI firms including OpenAI, Meta, Google and Anthropic warned automation of AI R&D may be near and, while timing is uncertain, a material risk exists that rapid capability breakthroughs could outpace human understanding and control. The signatories said industry, governments and society need time to address emerging risks, build safety guardrails and strengthen regulatory frameworks, but firms and countries face a competitive dilemma—no one

A joint statement signed by 1,132 employees at frontier AI firms including OpenAI, Meta, Google and Anthropic warned automation of AI R&D may be near and, while timing is uncertain, a material risk exists that rapid capability breakthroughs could outpace human understanding and control. The signatories said industry, governments and society need time to address emerging risks, build safety guardrails and strengthen regulatory frameworks, but firms and countries face a competitive dilemma—no one is willing to slow alone. They urged the US government to build on existing monitoring of frontier model releases and push an international collaboration to develop the technical and governance tools needed to prudently guide automated AI R&D.