AI workers urge US support for tools to pace automated AI development

More than 1,100 AI workers signed an open letter asking the US government to support an international effort to develop technical and governance tools that could deliberately pace frontier automated AI development. The signers include workers from companies such as OpenAI, Anthropic, Google, and Meta. The request is framed around preparing for a point when AI systems may become capable of developing themselves.

The letter focuses on automated AI development, described in the article as the point when AI can develop itself, also called recursive self-improvement. The concern is that such a shift could accelerate AI progress faster than people can understand or control the resulting systems. Anthropic has said its Claude models are quickly approaching that threshold.

Named signatories include Anthropic cofounders Jack Clark and Jared Kaplan, OpenAI chief scientist Jakub Pachocki, Meta chief scientist Shengjia Zhao, and Google DeepMind AI safety and alignment lead Anca Dragan. John Schulman, chief scientist at Thinking Machines, said he signed to highlight the need for coordination mechanisms as automated AI research accelerates progress. He also said labs should begin designing such mechanisms voluntarily before US government involvement.

The letter arrives as AI leaders and companies increasingly discuss global regulation for frontier models. Business Insider reports that Dario Amodei, Sam Altman, and Demis Hassabis have each outlined frameworks for AI regulation, though their details vary. The article also notes that Congress has not produced a federal policy, while states have begun imposing their own industry rules.

The core issue is not only whether AI should be regulated, but how any system could slow or review frontier-wide progress without putting one company or country at a competitive disadvantage. The letter says the world currently lacks the tools to deliberately pace that progress. For AI labs, the practical takeaway is to prepare coordination and review mechanisms before automated AI development becomes more advanced.