Chinese open source AI ban draws pushback from Nvidia, Meta and OpenAI

Major AI companies are opposing a possible Chinese open source AI ban as the Trump administration considers limits on access to advanced models from China, Fortune reported. The debate matters because Washington is weighing cybersecurity risks against an AI industry that says broad access helps innovation and U.S. influence.

Fortune reported that the administration has been considering the idea with a small group of advisers. The concern, according to Fortune, is that powerful Chinese models could aid cyberattacks or erode the U.S. lead in AI.

The discussion intensified after Beijing-based Moonshot AI released Kimi K3 this month, Fortune reported. The model is widely available to download for free and rivals leading U.S. models, according to Fortune.

Why do AI companies oppose a Chinese AI models ban?

Nvidia CEO Jensen Huang used his first post on X to argue against a potential ban, sharing an open letter called “Open Weights and American AI Leadership.” The letter was signed by companies including Meta, Perplexity and Microsoft; Microsoft later posted a version on its website with a longer list of signatories that included OpenAI, Fortune reported.

Huang wrote that open models strengthen safety and cybersecurity, speed innovation and help sovereignty. In this debate, open-weight or open models generally means AI systems whose underlying model files can be downloaded and used outside a single company-controlled service.

OpenAI president and co-founder Greg Brockman did not directly call for rejecting a ban when he met with journalists on July 23, Fortune reported. He said OpenAI believes in access and that having more models is beneficial, according to Fortune. Brockman became a major Trump donor in January 2026 with a $26 million contribution to the MAGA Inc. super PAC, SFGate reported, according to Fortune.

Replit also signed the letter. Michele Catasta, Replit’s president and head of AI, told Fortune that a ban would set a bad precedent and that Chinese models could put downward pressure on AI prices. He also said Kimi K3 had been “pretty exceptional” at design in Replit’s early testing.

What did Anthropic say about open models?

Anthropic did not sign the open letter. After pressure to state its view, CEO Dario Amodei wrote in a blog post that he does not support banning open source models, calling that approach protectionist and ineffective, Fortune reported.

Amodei also challenged one of the letter’s central safety claims. He wrote that he does not agree open-weight models necessarily make safeguards easier to build or that broad access to advanced capabilities necessarily helps defenders more than attackers. He said the opposite may be just as likely.

Amodei said he was more concerned about restricting chip exports to China, stopping Chinese companies from distilling U.S. models and requiring stronger safety testing for both open and closed models, Fortune reported.

Fortune reported that the companies opposing a ban also have business reasons to favor open models. Nvidia benefits when more groups need chips to run AI systems, while companies such as Replit can use open models for more control, customization and lower costs.

The debate also comes as Washington considers broader oversight of model releases. The Information reported that the Trump administration is finalizing a framework for AI companies to submit advanced models to the government before public release, according to Fortune.

This story draws on original reporting from Fortune.