Nvidia and peers oppose curbs on ‘open’ AI models

By the Editorial Team · Saturday, July 25, 2026

Leading US tech companies, including Nvidia and Microsoft, have urged the federal government not to restrict open-weight AI models amid concerns over intellectual property theft and Chinese AI advancements. They argue that maintaining open AI research is crucial for innovation and competition, even as policymakers consider sanctions in response to alleged covert technology transfer by Chinese firms.

Leading US technology companies have urged the federal government not to impose restrictions on "open-weight" artificial intelligence (AI) models, amid growing tensions over Chinese AI developments and concerns about intellectual property theft. Nvidia, Palantir, Microsoft, Meta, and others signed an open letter this week cautioning against premature curbs on open AI systems, emphasizing that maintaining diversity in AI research and development is essential for competition and innovation.

Open-weight models are AI systems whose underlying code and parameters can be freely accessed, modified, and run on various hardware, differentiating them from closed models like those produced by OpenAI, Anthropic, and Google. These open models have been particularly embraced by Chinese AI developers, who have recently unveiled notable advancements, such as the Kimi K3 bot developed by Beijing-based Moonshot AI. The Kimi K3 launch has intensified calls from some US policymakers to restrict or sanction Chinese AI labs, citing national security concerns and fears that China may be gaining ground in the global AI race.

Officials in Washington, including Michael Kratsios, director of the White House Office of Science and Technology Policy, have alleged that Moonshot employed a technique called "distillation" to derive its model from Anthropic’s Fable system. Distillation involves training a less powerful AI model based on the outputs of a more advanced one. Kratsios characterized this as "large-scale, covert industrial distillation aimed at stealing proprietary US technology," deeming it unacceptable. The US Treasury secretary, Scott Bessent, echoed these concerns, warning that sanctions could be considered if Chinese firms engage in covert intellectual property theft. Anthropic responded by highlighting the national security risks posed by industrial espionage.

However, the letter from US tech firms argued that banning or restricting open AI models could stifle innovation and limit access to affordable AI tools. The signatories stressed that open-weight AI technology helps lower costs for businesses and promotes a wider adoption of AI capabilities, contrasting with closed models that dominate much of Silicon Valley’s market. Nvidia CEO Jensen Huang publicly opposed potential restrictions, stating that the global AI ecosystem benefits from both open and closed frontier models.

Notably, OpenAI, Anthropic, and Google declined to join the letter, reflecting differing perspectives within the US tech industry. Some investors and executives have accused companies like Anthropic of amplifying the threat posed by open-source AI to consolidate their market positions.

Preliminary assessments of the Kimi K3’s abilities by US and UK researchers indicate that while it has made progress, it still lags behind the most advanced American AI models. Moreover, US-developed open-weight models were found to be less powerful still.

As US policymakers weigh responses to China's recent AI advances, the debate continues over how best to balance national security concerns with fostering an open, competitive environment for AI development domestically and internationally.

This article was reported by synthesizing coverage from multiple independent news outlets to present a balanced, unbiased account. Read about our editorial standards. Spotted an error? Let us know.