Mark Zuckerberg, CEO of Meta, has voiced his opposition to the U.S. blocking Chinese AI models, arguing that such a move would not provide an effective solution in the ongoing technological competition. This was reported by the Financial Times.

Zuckerberg emphasized that American companies should focus on identifying their own weaknesses and barriers to enhance their competitive edge against Chinese developers.

Debate Intensifies Over Kimi K3

This statement comes amid discussions in Washington regarding potential restrictions on Chinese open-weight models. According to Reuters, the conversation gained momentum following the release of Kimi K3 by Beijing's Moonshot AI.

Developers touted it as the world's first open model of the 3T class, featuring 2.8 trillion parameters, native vision, and a context span of 1 million tokens. The project claims it ranks just behind proprietary models Claude Fable 5 and GPT 5.6 Sol.

Some U.S. officials and companies suspect that Chinese labs may have engaged in model distillation. In July, Michael Kratsios, Director of the White House Office of Science and Technology Policy, accused Moonshot AI of this practice. That same month, U.S. Treasury Secretary Scott Bessent noted that the government would be scrutinizing Chinese open-source models for signs of intellectual property theft.

He later clarified that sanctions and inclusion on export restriction lists could be imposed for "covert industrial distillation attacks."

Industry Defends Open Weights

Zuckerberg's stance aligns with a broader campaign advocating for open-weight models. The Linux Foundation announced its support for a letter signed by Cisco, Dell Technologies, IBM, Meta, Microsoft, Nvidia, and others.

As AI integrates across the global economy, how do we ensure it’s safe, transparent, and trustworthy?

At the Linux Foundation, we believe the answer is openness. Open-weight models give researchers & enterprises the visibility needed to inspect, audit, and fix vulnerabilities… pic.twitter.com/KD6Gr0sfdm

— The Linux Foundation (@linuxfoundation) July 27, 2026

“As AI integrates into the global economy, how can we ensure its safety, transparency, and reliability? At the Linux Foundation, we believe the answer lies in openness. Open-weight models provide the necessary visibility for researchers and enterprises to inspect, audit, and address vulnerabilities before they can be exploited,” the post stated.

Nvidia CEO Jensen Huang also opposed the bans, stating in an interview with Axios that beneficial Chinese open models should be utilized, warning that blocking them could ultimately harm the U.S.

Meanwhile, the White House is establishing a mechanism for pre-screening the most powerful AI systems. In a June directive, authorities were tasked with developing a voluntary framework that would allow developers to give the government access to cutting-edge models up to 30 days prior to their release.

In July, participants in a march in San Francisco called for OpenAI, Anthropic, and Google DeepMind to coordinate a halt on training more powerful models, suggesting that the resources freed up be redirected towards safety.

Also in July, former OpenAI researcher Daniel Kokotailo proposed pausing the AI race until 2040.