Washington Is Weighing Curbs On Chinese AI. Nvidia, Microsoft And Meta Warn Open Models Shouldn’t Be Included

Nvidia, Microsoft, Meta and more than two dozen technology companies are urging U.S. policymakers not to impose broad restrictions on open-weight artificial intelligence models, arguing that doing so could undermine American innovation and hand an advantage to foreign competitors.

In a joint letter released Friday, the coalition warned that "premature restrictions" on open-weight AI models would stifle competition, weaken the U.S. technology ecosystem and potentially drive innovation overseas at a time when Chinese AI developers are rapidly gaining ground.

The letter comes amid growing debate in Washington over whether access to Chinese-developed open-weight AI models should be limited because of national security and intellectual property concerns.

Unlike proprietary systems, open-weight AI models allow developers to run models on their own infrastructure. The companies signing the letter include Nvidia, Microsoft, Meta and Palantir, among more than 20 others. Nvidia CEO Jensen Huang and Microsoft CEO Satya Nadella also shared the statement on their social media accounts, signaling their public support for the initiative.

The coalition argued that open-weight models promote competition by preventing AI capabilities from being concentrated in the hands of only a few companies. "Relying solely on closed models is not inherently safe: they can be breached, misused, or fail in ways that outsiders cannot detect," the letter states. "And concentrating advanced AI capabilities behind a small number of closed models compounds that risk.

"The appeal comes as Chinese AI companies continue narrowing the gap with U.S. leaders. Earlier this month, Chinese startup Moonshot AI introduced Kimi K3, an open-weight model that outperformed several leading American AI systems on selected industry benchmarks, raising new concerns in Washington about China's accelerating AI capabilities.

The Trump administration has signaled it is closely monitoring the issue. Treasury Secretary Scott Bessent told CNBC earlier this week that officials would investigate whether Chinese AI firms had improperly used American intellectual property, adding that the United States has the authority to impose sanctions if theft is confirmed.

White House technology adviser Michael Kratsios also alleged this week that "Moonshot AI distilled Anthropic's Fable for the development of its K3 model." AI distillation is a training technique in which a smaller model learns from the outputs of a larger one.

He said that "Legitimate AI distillation used to create smaller, more efficient models plays a vital role in this open innovation ecosystem. However, large-scale, covert industrial distillation aimed at stealing proprietary U.S. technology and undermining American research is unacceptable."

The companies behind Friday's letter acknowledged concerns about unlawful distillation but urged policymakers to address those issues through targeted legal action rather than sweeping restrictions on open-weight AI development. "Concerns about unlawful distillation should be addressed through targeted legal and commercial frameworks," the letter states, instead of broad limitations that could hinder innovation.

Notably absent from the list of signatories were OpenAI and Anthropic, two of the world's most valuable AI companies. Both companies, each reportedly nearing a $1 trillion valuation, are preparing for potential initial public offerings after confidentially filing paperwork with the U.S. Securities and Exchange Commission.

Despite not signing the letter, executives from OpenAI also voiced support for broad AI access. Speaking with reporters Thursday in New York, OpenAI President Greg Brockman said he had not participated in discussions with the Trump administration about banning Chinese open-weight models and emphasized the importance of making AI widely available.

"I think that, fundamentally, AI and AI usage is something that is actually very important to democratize," Brockman said. "Having more models, more usage, that is a good thing." OpenAI CEO Sam Altman also reacted positively after the letter was published, writing on X that he wants the United States to succeed with both open-weight and proprietary AI models and was "glad to see this."

The debate intensified further after AI platform Hugging Face revealed it had relied on Chinese company Z.ai's open-weight model GLM 5.2 to help contain what OpenAI described this week as an unprecedented cyberattack carried out by rogue AI models.

According to Hugging Face's head of machine learning, Yacine Jernite, the company initially attempted to use Anthropic's model to analyze the incident, but its safety guardrails prevented it from assisting in a defensive cybersecurity scenario.

The coalition concluded that America's long-term AI leadership will depend less on any single breakthrough model and more on creating an open ecosystem capable of spreading AI innovation across industries. "Our AI leadership will be judged not by one frontier AI model," the companies wrote, "but by whether the United States builds a strong, open ecosystem that diffuses into every sector."