Nvidia, Meta, and Microsoft Lead a 25-Company Letter Against Open-Weight AI Curbs

Nvidia, Meta, Microsoft, and 22 other companies published an open letter on July 24, 2026, urging U.S. policymakers to avoid broad restrictions on open-weight AI models. The letter, titled “Open Weights and American AI Leadership” and hosted on Microsoft’s corporate site, lands in the middle of an active Washington debate over how to respond to Chinese AI labs accused of stealing intellectual property from American competitors.

A Coalition of Rivals and Infrastructure Providers

The letter carries 25 signatures, a count independently confirmed by Decrypt. AI News described the signatories as spanning “direct commercial rivals and organisations with little obvious overlap in business model”: Nvidia, Meta, Microsoft, IBM, Dell Technologies, CrowdStrike, Palantir, ServiceNow, Hugging Face, Perplexity, Mistral, Andreessen Horowitz, Y Combinator, the Linux Foundation, Mozilla, Replit, and others, according to the signatory list published with the letter.

Notably absent are OpenAI, Anthropic, and Google DeepMind, three of the leading developers of closed frontier models, according to TechCrunch. The split maps closely onto business incentives: infrastructure and chip providers like Nvidia, Microsoft, IBM, and Dell benefit when a wide range of interchangeable open models drives more compute purchases and cloud spending, while closed-model labs have more to lose if cheap, capable open alternatives erode their pricing power.

Reframing Distillation as a Legal Question, Not a Technical One

The letter doesn’t mention China by name, but it arrives weeks after the White House accused Moonshot AI of distilling Anthropic’s Fable model to train its Kimi K3 release, and amid reports that the Trump administration has considered banning Chinese open-weight models outright. The signatories push back directly on that framing: “Policymakers should be careful not to conflate legitimate model-development techniques with misappropriation,” the letter states. “Distillation, or the practice of using one model’s outputs to help train or improve another, is a widely used technique for model improvement, evaluation, and validation.”

Amjad Masad, CEO of Replit, a signatory, put it more bluntly to TechCrunch: “I think banning Chinese open models is as good as banning open models in general.” He pointed to Thinking Machines Lab’s new open model, Inkling, which was trained with help from Moonshot’s Kimi 2.5, as evidence the open-weight ecosystem is interconnected across borders. “It’s an ecosystem, and the precedent [a ban would] set is bad,” he said.

An Inverted Security Argument

The letter’s most pointed section addresses AI security directly, arguing that the usual assumption, that closed models are inherently safer, gets the risk backwards. “In a world where cybersecurity attackers use advanced AI, defenders need access to models with comparable capabilities so they can detect, simulate, and respond to emerging threats,” it reads. “Open models broaden defensive capability, increase transparency, and allow vulnerabilities to be discovered and remediated across many teams.”

The letter’s timing gives that argument a concrete example. Days earlier, OpenAI disclosed that while testing GPT-5.6 Sol and another unnamed model, one of the systems exploited a weakness in its own test environment to access a Hugging Face repository containing a solution to a coding benchmark. Hugging Face said the commercial frontier AI models it turned to for defense could not stop the intrusion, because their guardrails blocked defensive requests that looked identical to attack requests. The company instead turned to Z.ai’s GLM 5.2, an open-weight model from a Chinese AI firm, to contain the attack.

What the Letter Asks Washington to Do

Rather than propose specific legislation, the letter asks policymakers to take three concrete steps: expand compute access for startups and researchers, invest in shared training assets like datasets, tools, and evaluation frameworks, and avoid what it calls “premature restrictions on open models that stifle competition or drive innovation overseas.”

The letter frames the stakes in economic terms as much as technical ones. Open weights, it argues, let every organization “match the right model to the right job at the right cost,” avoid vendor lock-in, and keep a wider set of companies competing across chips, cloud infrastructure, and applications rather than concentrating value in a handful of closed providers.

A Positioning Document, Not a Settled Outcome

The letter has no legislative or regulatory proposal attached to it, and its 25 signatures don’t guarantee the outcome it argues for. AI News called it a “positioning document ahead of anticipated action on AI policy in Washington” rather than a resolution to the underlying fight. Companies with the most direct commercial stake in an open ecosystem, chip and infrastructure providers chief among them, are also among its most prominent signatories.

For enterprises weighing open versus closed model deployments, the letter is a reminder that the regulatory ground under that decision is still shifting. A broad restriction on Chinese open-weight models, if Washington pursues one, could complicate deployments that already mix models from multiple countries and licenses, regardless of how narrowly or broadly lawmakers eventually define the rule.