Open-weight AI restrictions draw warning from Nvidia, Microsoft and Meta

Nvidia, Microsoft, Meta, Palantir and other tech companies are pushing back on open-weight AI restrictions as Washington weighs how to respond to fast-improving Chinese models. For everyday investors, the fight matters because AI policy can shape who gets to build, use and profit from the technology, from chipmakers to software platforms to IPO-bound startups.

A group of 25 tech companies released a letter Friday urging policymakers to avoid “premature restrictions” on open-weight artificial intelligence models. The companies warned that broad limits could “stifle competition or drive innovation overseas,” according to the letter.

What are open-weight AI models?

Open-weight AI models are systems whose underlying model weights are made available so users can download, change and run them on their own computing infrastructure. That differs from closed models, where the developer keeps the model controlled behind its own services or interfaces.

The distinction has become a live policy issue because open-weight models can spread faster and be adapted by more users. Supporters say that helps competition and access. Critics and government officials have raised concerns about security, misuse and whether some foreign models rely on U.S. technology without permission.

Why are companies warning about open-weight AI restrictions?

In their letter, the companies argued that open-weight models help keep AI benefits from being concentrated among a small number of closed-model providers. They also wrote that relying only on closed systems does not eliminate risk, because closed models can still be compromised, misused or fail in ways outside observers cannot see.

Nvidia CEO Jensen Huang and Microsoft CEO Satya Nadella both shared the letter on their personal social media accounts, according to CNBC. Meta was also among the companies backing the warning, while OpenAI and Anthropic did not sign.

OpenAI and Anthropic primarily develop proprietary, closed AI models, according to CNBC. Both companies are valued at nearly $1 trillion and are preparing for potentially large initial public offerings that could come as soon as this year, CNBC reported. Anthropic confidentially filed its prospectus with the Securities and Exchange Commission in June, and OpenAI followed days later, according to CNBC.

Why Chinese AI models are part of the debate

The policy pressure has grown as Chinese open-weight models gain ground against leading U.S. offerings. CNBC reported that Chinese startup Moonshot AI drew attention this month after releasing Kimi K3, a model that beat advanced American models on some industry benchmarks.

U.S. Treasury Secretary Scott Bessent told CNBC on Tuesday that the Trump administration would examine whether Chinese companies were stealing American intellectual property. Bessent said the government has “the ability to sanction them because of this theft.”

White House advisor Michael Kratsios wrote Wednesday on X that Moonshot AI developed Kimi K3 by distilling Anthropic’s technology. Distillation is an AI training method in which a smaller model is trained using the outputs of a stronger existing model.

Kratsios wrote that legitimate distillation is important to open innovation, while calling “large-scale, covert industrial distillation aimed at stealing proprietary U.S. technology” unacceptable. The companies’ Friday letter said concerns about unlawful distillation should be handled through “targeted legal and commercial frameworks,” rather than broad limits on techniques they said are important to AI development.

OpenAI President Greg Brockman told reporters Thursday in New York that he believes AI access should be broad and said he had not been part of any Trump administration talks about banning Chinese open-weight models in the U.S. “I think that having more models, more usage, that is a good thing,” Brockman said.

This story draws on original reporting from CNBC.