Both Labs Skipped the Letter. One Eventually Signed. Follow the Money.

The letter was three pages. The title was "Open Weights and American AI Leadership." Twenty-five companies signed it on July 24, including Nvidia, Microsoft, Meta, IBM, Palantir, Hugging Face, Mozilla, Andreessen Horowitz, and Y Combinator. The ask was simple: do not impose premature restrictions on open-weight AI models.

OpenAI and Anthropic were not on it.

OpenAI eventually added its signature after the omission was widely noticed, and Sam Altman reposted a note on X in support of open-source models. Anthropic held out. Dario Amodei has described open source as "a red herring," saying he does not think it works the same way in AI as in other areas.

The consensus read is that both labs are cautious but not opposed. That read is wrong, and expensive to believe.

The Lobbying Numbers Say Something Else

OpenAI and Anthropic spent a combined $3.17 million on lobbying during Q2 2026, a 23% jump from the previous quarter, according to federal disclosure filings. Anthropic spent nearly $2 million on lobbying in Q2, focusing on export controls, cybersecurity, and AI safety standards. Lobbyists met with lawmakers in both chambers of Congress, as well as officials at the White House, Commerce Department, and Treasury Department.

That is not the lobbying schedule of a company that finds open-source debate to be a red herring.

OpenAI and Anthropic compete fiercely for customers. In Washington, the two leading AI labs are aligning on one issue: warning policymakers about the risks of powerful Chinese open-weight AI models. Closed-model labs stand to benefit from greater scrutiny of open-weight models.

Translation: we think open source is irrelevant, and we are spending millions of dollars making sure regulators agree with us.

Just five weeks ago, Anthropic proposed giving the government standing authority to block frontier model releases that fail independent safety tests, with penalties tied to global revenue. "Frontier AI models, like airplanes, should be required to go through technical testing and auditing, and their release should be blocked or reversed as a threat to public safety if they do not meet high standards of safety," Amodei wrote.

That framing sounds like safety. It functions like a licensing regime. Only the largest labs have the compliance infrastructure to survive one.

The Friday Signature Is Damage Control

OpenAI added its signature only after the omission was widely noticed on X. Sam Altman's post said he wants the US to win in both open-source and proprietary AI. That sentence does not say OpenAI opposes restrictions. It says they want to win.

Anthropic hasn't just declined to champion open weights — it has actively lobbied for restrictions on the open-weight models coming out of China. The company's position has been consistent since 2023. In July 2023, Amodei appeared before the US Senate Judiciary Committee and told Congress that open-source AI was moving down a "very dangerous path." Only the lobbying budget has scaled.

The Economist reported that David Sacks, an adviser to the Trump administration on AI, accused Anthropic of running a "sophisticated regulatory-capture strategy based on fearmongering." Sacks said the leading AI labs were hoping to create complicated safety rules that only they would comply with, making it harder for younger startups to challenge their market position.

When the Trump administration's own AI czar says it out loud in the press, the deniability window is closing.

The economic logic is straightforward. At Nvidia's CES 2026 press Q&A, Jensen Huang said one in every four tokens generated today comes from an open model. Weights that anyone can download get served from enterprise clusters, regional clouds, and on-premises racks rather than a handful of hyperscaler API endpoints. If open-source development gets regulated to the point of friction — licensing, mandatory audits, compute caps framed as safety standards — those workloads migrate back to closed APIs. Both labs know exactly what that migration is worth.

Both companies, valued at nearly $1 trillion each, are gearing up for potentially massive initial public offerings that could land as soon as this year. Regulatory moats look prettier on an S-1 than product moats. Product moats require you to keep winning. Regulatory moats require you to keep lobbying.

The forward question is whether the open-weights coalition can convert a signed letter into policy cover before any distillation or export rule is drafted. That is the moment the 25 signatories — and every startup building on open weights — actually need it.

If you are filing your own position with a regulator, building on open-weight models, or making inference infrastructure bets: the two largest AI labs are playing for a specific outcome in Washington, and it is not the one their press statements describe.

What to watch: Whether regulatory language emerges that mirrors Anthropic's framing. Whether OpenAI's next policy paper calls for licensing or audit requirements. Whether the coalition signatories formalize into an amicus bloc in upcoming rulemaking. Whether Anthropic's Q3 lobbying disclosure shows the spend increasing again.