OpenAI, along with roughly a thousand employees from other labs — including Google, Anthropic, and Meta — signed the "Pacing the Frontier" letter.
What the letter says
> Artificial intelligence could help build a dramatically better future, but that outcome isn't guaranteed. Leading AI companies believe they may be close to automating AI research itself. It's hard to predict exactly how much this will accelerate AI progress, but there's a real risk that capabilities will advance faster than our ability to understand and control the systems being built.
>
> To realize AI's potential, industry, government, and society at large may need the ability to buy time to address emerging risks, develop safety measures, and strengthen oversight. But every company — like every country — faces intense competitive pressure that makes unilaterally slowing down untenable. And right now, the world lacks the technical and governance tools to deliberately regulate the pace of global progress in frontier AI development.
>
> Building on existing work to monitor frontier AI model releases, we call on the US government to:
> - support international efforts to develop the technical and governance tools needed to deliberately regulate the pace of progress at the frontier of automated AI development
What's next
On August 1st, the US is due to define what counts as a "frontier system" and which models fall under that label. This matters because frontier systems will be subject to new pre-release evaluation rules — at minimum, the government will need access to a model 30 days before its public release.
Government AI regulation is already showing up elsewhere: South Korea recently passed the world's first comprehensive AI law.