A coalition of 25 US companies, including Nvidia, is urging the White House not to impose broad restrictions on open-weight AI models as Washington weighs its response to Chinese AI systems on “national security” grounds.
In an open letter released on Friday, the group warned against “premature restrictions” on open-weight models, drawing parallels with fears over open-source software in the 1980s.
It argued that AI leadership will be determined not by a single frontier model, but by the ability to build “a strong, open ecosystem that diffuses into every sector”.
The signatories described open-weight models — whose core components can be downloaded, inspected, modified and run on independent infrastructure — as an important foundation because they widen access to advanced AI and make it more adaptable.
In his first-ever post on X, Nvidia CEO Jensen Huang shared the letter and wrote: “AI will transform every industry, power every company, and be built by every country. Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.”
The world needs both frontier closed models and frontier open models, he added.
Although the letter does not mention China directly, it comes as Washington debates how to respond to allegations that Chinese AI labs have been extracting intellectual property from United States competitors while rapidly closing the capability gap.
US Treasury Secretary Scott Bessent said on Wednesday that sanctions on Chinese AI companies remain under consideration. His remark followed accusations by White House science and technology policy chief Michael Kratsios that Chinese startup Moonshot AI had distilled the Fable model developed by US firm Anthropic.
Distillation is a process in which an AI model is trained to mimic the output of a larger, more capable one.
On Monday, Beijing urged Washington to stop smearing Chinese AI companies and threatening them with sanctions, saying such actions lack factual and legal grounds and constitute a typical form of AI hegemony.
“China will take all necessary measures to firmly safeguard its legitimate and lawful rights and interests,” the Commerce Ministry said in a statement.
Moonshot AI unveiled its latest open model, Kimi K3, on July 16. With 2.8 trillion parameters, it is reportedly the largest open-weight AI system by that measure and is said to approach the performance of Anthropic’s Claude and OpenAI’s ChatGPT.
Some researchers have questioned whether Kimi K3 could have been built primarily through distillation from Fable, which has been publicly available only since July 1.
Braden Hancock, a researcher at the Laude Institute and cofounder of Snorkel AI, told online newspaper TechCrunch that there was simply not enough time to distill that volume of data, train a model and release it within two weeks.
Nathan Lambert, an AI researcher at the Allen Institute for AI, said in a podcast released on Wednesday that Washington’s consideration of restrictions on top Chinese open-weight models is “downstream of some fear-mongering”.
As Chinese models near the frontier and shift their training toward reinforcement learning, distillation is becoming less consequential, he said.
The letter addressed the distillation debate, describing the practice as a long-standing method for improving, evaluating and validating AI models, rooted in the open-source software tradition of building on existing technology.
- The impact of AI tech without a manual in Africa
- CEOs bracing for threat to jobs from AI
- Taking advantage of AI to reduce gap in tech race
While illegally extracting value from closed models is a legitimate concern, it should be addressed through targeted legal and commercial frameworks rather than broad restrictions on the techniques, the letter said.
“Banning Chinese open models is as good as banning open models in general,” Replit CEO Amjad Masad, whose company also signed the letter, told TechCrunch.
Masad said Thinking Machines Lab’s new open model, Inkling, was trained with the help of Moonshot’s Kimi 2.5. “It’s an ecosystem, and the precedent (a ban would) set is bad,” he said.
The letter also challenged the notion that open-weight models are inherently dangerous because they widen access to powerful systems that could be misused in cyberattacks.
ALSO READ: AI development should be innovative, inclusive and cooperative, not zero-sum
“In fact, openness may be one of the most important paths to AI safety and security,” it said. “Relying solely on closed models is not inherently safe: they can be breached, misused, or fail in ways that outsiders cannot detect.”
Clement Delangue, CEO of Hugging Face, described the cybersecurity debate on open-source AI as “backward”.
In an X post on July 20, he wrote, “Open models aren’t the risk, they’re the defense,” referring to a case in which Chinese AI company Z.ai’s open model helped where leading frontier models had failed.
The letter reflects a divide within the US AI sector. Closed-model developers Anthropic and OpenAI have pressed the administration to restrict Chinese open-weight models as their business models are increasingly challenged by cost-effective, high-performing and widely accessible alternatives.
– China Daily
