Analysis
OpenAI and Anthropic have formally endorsed a letter, originally signed by more than 1,100 employees across frontier AI labs, calling on the US government to help internationally coordinate a deliberate pace for AI capability development. The letter, dubbed "Pacing the Frontier," started as a grassroots employee petition and was endorsed by both companies as organizations within hours of its publication -- an unusually fast institutional response to a bottom-up staff letter.
The list of signatories is notable for its seniority: Anthropic CEO Dario Amodei, OpenAI chief scientist Jakub Pachocki, OpenAI chief research officer Mark Chen, Meta AI chief scientist Shengjia Zhao, and Google DeepMind's VP of AI safety and alignment Anca Dragan all put their names on it. Getting that group to agree on anything publicly, given how competitive the frontier-lab race has become, is itself the story -- these are people who spend most of their public statements emphasizing why their lab is moving fastest, not why the whole industry should slow down together.
The letter's ask is specifically international: it wants US government support for diplomatic and governance mechanisms that could pace frontier development globally, not just domestically, on the theory that unilateral US restraint without a matching commitment from China or other labs would simply cede ground rather than reduce risk. It warns of a "real risk" that AI capability outpaces humanity's ability to understand or control it -- language pulled almost directly from years of AI-safety research now showing up in a jointly-endorsed corporate position.
The timing is pointed. This lands the same week Anthropic separately made news for being the only frontier lab to decline signing a different, Nvidia-led letter urging Washington not to restrict open-weight AI models -- a category where China currently leads. Amodei has publicly clarified Anthropic isn't advocating an outright ban on open-weight models, but the company's refusal to sign alongside Nvidia, OpenAI and others has drawn public criticism from White House AI adviser David Sacks, who suggested Anthropic is using safety rhetoric to protect its own commercial position.
Taken together, the two letters show an industry that agrees on wanting government involvement in principle but disagrees sharply on what that involvement should look like -- pacing capability development is one ask; restricting open-weight competition is a different one entirely, and Anthropic is now visibly isolated on the second even as it's aligned with rivals on the first.
For founders and investors, the practical read is that AI policy is not moving toward a single predictable regulatory outcome -- it's fragmenting into overlapping, sometimes contradictory asks from the labs that will most shape whatever rules eventually get written, which means startups building on frontier APIs should expect policy uncertainty to persist well into 2027 regardless of which letter gets more traction in Washington.
What to watch: whether the Trump administration responds substantively to the pacing letter given its parallel push (revealed the same week) to secure early government access to frontier models before public release, and whether Anthropic's isolation on the open-weight question hardens into a lasting rift with the rest of the industry.
