Nvidia, Microsoft, Meta and dozens of tech firms urge Washington not to restrict open-weight AI
On July 24, 2026, Nvidia, Microsoft, Meta, and 47 other technology companies, venture-capital firms, and nonprofits signed an open letter titled “Open Weights and American AI Leadership,” urging Washington not to restrict open-weight AI models and arguing that open weights are central to U.S. AI leadership. Nvidia CEO Jensen Huang first shared the letter on X in what was his first post on the platform, and within a day the list of signatories had roughly doubled as additional major companies joined.
Who signed the letter
Nvidia CEO Jensen Huang posted the letter on X with 25 initial signatories. Several prominent AI developers were not on the original list, including OpenAI. Within a day, the coalition had grown to around 50 signatories, with OpenAI, Google, AMD, Cisco, and others added. Anthropic did not sign. The published letter frames the coalition as a broad cross-section of the U.S. AI industry, spanning chipmakers, cloud providers, model developers, and investors.
What the letter argues
Open-weight models are AI models that anyone can download, inspect, modify, and run on their own infrastructure. The letter makes several claims about their value to the U.S. economy and to AI safety:
- Open weights expand access to the AI economy. Startups, businesses, universities, and public institutions can build on advanced models without training one from scratch or paying frontier-model prices for every task, and can pick a model that fits their budget and needs.
- The letter says open weights strengthen competition across model developers, cloud providers, chip makers, and application and service providers, keeping AI gains more widely shared and lowering costs.
- Open weights let organizations run models on their own infrastructure.
- The letter acknowledges open weights carry real risks. Once released, model weights cannot be recalled, and modified versions can be hard to trace. The signatories argue that the answer is not prohibition, because defenders need models with capabilities comparable to those used by attackers, and open models allow many teams to find and fix vulnerabilities.
- The letter argues that openness is a crucial path to AI safety and security, citing two claims without supporting evidence: that closed models are not inherently safe, and that concentrating advanced capability behind a few closed models creates single points of failure.
What the letter asks policymakers to do
The letter makes three concrete policy asks:
- Expand access to compute for startups and researchers.
- Invest in shared training assets such as datasets, tools, and evaluation frameworks.
- Keep the frontier plural by avoiding premature restrictions on emerging technologies.
It also warns policymakers not to conflate legitimate model-development techniques with misappropriation, stating that distillation, the practice of using one model’s outputs to train another, should not be confused with misappropriation.
Context and notable absences
The letter was published amid allegations that Chinese labs extracted intelligence from U.S. models, including Anthropic’s Fable. Washington’s response to those allegations remains unclear. Anthropic is the lone U.S. frontier AI lab not on the list of signatories.
The full letter is available as a PDF on Nvidia’s site.
FAQ
What is the “Open Weights and American AI Leadership” letter?
It is an open letter signed on July 24, 2026, by Nvidia, Microsoft, Meta, and 47 other companies, venture-capital firms, and nonprofits, urging Washington not to restrict open-weight AI models.
Which major AI companies signed the letter?
Nvidia, Microsoft, and Meta were among the initial 25 signatories. OpenAI, Google, AMD, and Cisco joined within a day, bringing the total to around 50. Anthropic did not sign.
Why are the signatories asking Washington to keep open-weight AI available?
The letter argues that open-weight models broaden access to advanced AI, boost competition, give users more control, and help defenders keep pace with attackers, and that restricting them could push innovation abroad.
Related coverage
- Poolside releases Laguna S 2.1, an open-weight coding model that beats rivals 10x its size
- Introducing Antares: Efficient Open-Weight Models for Vulnerability Localization
Related coverage
- Poolside releases Laguna S 2.1, an open-weight coding model that beats rivals 10x its size
- Introducing Antares: Efficient Open-Weight Models for Vulnerability Localization