Anthropic is the only major AI company refusing to sign a letter defending open-weights artificial intelligence, even as Nvidia, Meta, Microsoft, OpenAI, Google, and SpaceX have all publicly backed the movement. The company's isolation in the debate has sparked criticism from venture capitalists and industry leaders, but Anthropic's position reflects a deeper tension between safety arguments and commercial interests that the company has yet to address publicly.

What Happened When Anthropic's Most Powerful Models Went Dark?

On June 12, 2026, the U.S. government ordered Anthropic to restrict access to its Fable 5 and Mythos 5 models, barring foreign nationals from using them. Because enforcing a nationality rule across a shared cloud service proved impractical, Anthropic shut down both models worldwide. The company called the order "disproportionate," arguing that the capability behind the restriction was already widely available from competitors like OpenAI's GPT-5.5.

Anthropic warned that the precedent could "halt all new model deployments for all frontier model providers." The company was not alone in its criticism. Within three days, roughly 100 cybersecurity leaders, including former Facebook security chief Alex Stamos and executives from Luta Security, SocialProof Security, Veracode, and Sophos, signed an open letter demanding the ban be reversed. They argued the action "has taken the best models away from defenders, created market uncertainty, and risked America's AI leadership without any real risk to justify it".

Proof Security, Veracode, and Sophos, signed an open letter demanding the ban be reversed

Why Is Anthropic's Silence on Open Weights So Controversial?

On Friday, July 25, 2026, 25 companies signed "Open Weights and American AI Leadership," a letter defending open-source AI models. By Monday, July 28, two additional major companies had joined: OpenAI and Google. That left Anthropic as the only significant holdout among the tech industry's largest players.

The company's refusal to sign has drawn sharp criticism. David Sacks, a venture capitalist and former AI adviser to the Trump administration, posted bluntly on X: "The entire tech industry (save for Anthropic) has come out in favor of open source AI. They won't stop until they kneecap open source. The rest of the industry needs to watch these guys like a hawk." Bill Gurley, a general partner at Benchmark, interpreted Anthropic's reluctance differently, suggesting it reflects that open weights compete "with their corporate economic strategy." Kai-Fu Lee, founder of the Chinese open-model startup 01.AI, noted: "Who DIDN'T sign the Open Weight letter is far more interesting than who did," tagging Claude, Anthropic's flagship model.

What Does Anthropic Actually Believe About Open-Weights Models?

Anthropic's position on open-weights models has remained consistent for years. CEO Dario Amodei has argued that developers lose control of a frontier model the moment its weights are published. Once a file is released, thousands of people can download it, and there is no way to retrieve it. Amodei has never claimed frontier models should be unrestricted; instead, he has argued they should remain in the hands of a small number of accountable labs.

Anthropic's alternative to releasing model weights is gated access. The company launched Project Glasswing to give vetted organizations early access to its cyber models so they can find and fix security flaws without receiving the underlying weights. This approach allows the company to maintain control while still supporting security research.

How Does Anthropic's Position Compare to Its June Arguments?

The contradiction at the heart of this story is striking. In June, Anthropic argued that restricting one category of capable model does not remove the capability; it removes it from people who follow the rules. The company made this case against the government's export-control measure on Fable 5 and Mythos 5. In July, the open-weights letter makes nearly the same argument: restricting Chinese open-weights models would take the best tools away from defenders without meaningfully reducing capability for adversaries.

Yet Anthropic has not drawn this distinction publicly. The company did not respond to requests for comment from Business Insider. Until Anthropic explains why the same logic applies differently to government restrictions on its own models versus industry restrictions on Chinese models, the silence itself becomes the message.

What Commercial Interests Might Be Shaping Anthropic's Position?

Anthropic has one additional reason to support restrictions on Chinese open-weights models that goes beyond safety arguments. The White House has accused Moonshot, a Chinese AI company, of distilling Anthropic's Fable model to build Kimi K3, an open-weights release that triggered the current debate. Treasury Secretary Scott Bessent has floated sanctions in response. Of every company in this argument, Anthropic is the one with a direct commercial injury to point at.

This creates a potential conflict of interest. A company that spent June insisting restrictions punish defenders rather than adversaries has spent July saying nothing at all about restrictions that would benefit its own competitive position. The company wanted its Fable 5 model kept on sale; it did not ask to give the weights away. Selling access to a model you continue to control is fundamentally different from publishing one that anyone can copy and modify forever.

Steps to Understanding Anthropic's Strategic Position

• Safety vs. Commerce: Anthropic's stated position emphasizes keeping frontier models in the hands of accountable labs, which is a genuine safety argument but also happens to protect the company's business model of selling controlled access.
• The Export Control Precedent: When the U.S. government restricted Fable 5 and Mythos 5, Anthropic argued the measure was disproportionate and would harm defenders. The same logic could apply to restrictions on Chinese models, yet the company has not made this argument publicly.
• Commercial Injury: Anthropic has alleged that Moonshot distilled its Fable model to create Kimi K3, giving the company a direct financial stake in restricting Chinese open-weights models beyond abstract safety concerns.

Anthropic's position is internally coherent. Amodei has never claimed frontier models should be unrestricted; he has consistently argued they should stay in the hands of a small number of accountable labs. That argument works against both the open-weights letter and the June government order. But as venture capitalist Bill Gurley noted, it is also "extremely convenient" for a company whose business model depends on maintaining control over its most powerful models.

The real issue is not whether Anthropic's safety case is wrong. It is that the company has not publicly distinguished between restricting its own models for national security reasons and restricting competitors' models for the same stated reasons. Until Anthropic draws that distinction, the silence will continue to invite the interpretation that commercial interests, not just safety principles, are driving the company's refusal to sign the open-weights letter.