Trump Administration Reportedly Reviving Push to Ban Chinese AI Models After Kimi K3 Launch

A renewed U.S. effort to restrict Chinese open-weight AI models could affect how businesses buy, self-host and present their AI services.

Updated November 4, 2026 at 9:14 PM ET

Businesses using Chinese open-weight AI models now face a clearer procurement risk, as the Trump administration is reportedly reviving efforts to ban leading Chinese AI models following Moonshot AI’s release of Kimi K3. The renewed push centers on cybersecurity concerns, but enforcement against downloadable model weights would be difficult and could reshape AI costs, hosting decisions and vendor trust for U.S. companies.

Federal agencies already have a specific procurement restriction in place. The Congressional Research Service reported that “federal departments and agencies are required to bar DeepSeek, its affiliated entities, and their products from procurements.” The order also requires previously acquired products to be removed and extends to federal contractors and firms doing business with a prime contractor, according to the Congressional Research Service.

Why the Restrictions Are Returning

According to an Axios report dated July 20, the administration is reigniting earlier attempts to curb the growth of Chinese AI models in the United States. The renewed effort followed the release of Kimi K3, a model from the Chinese startup Moonshot AI.

The Department of Commerce previously considered adding several Chinese AI laboratories, including DeepSeek, to the Entity List. The list is maintained by the Bureau of Industry and Security and restricts access to sensitive American hardware, software and technology.

Officials also considered a joint advisory from the National Security Agency and the Office of the National Cyber Director to discourage the use of Chinese AI models. Another draft would have held U.S. companies liable for security breaches involving hosted Chinese models. Those proposals were paused because of internal concern about their market effects before being revived.

Why Businesses Are Adopting Chinese Models

Chinese models such as DeepSeek V4 and Kimi K3 publish trained weights for public download. Companies can run them on private infrastructure, keep data in-house and reduce inference costs compared with closed Western alternatives.

The price difference is substantial. DeepSeek-V4-Pro charges $0.87 per million output tokens, compared with $50 for Anthropic’s frontier Claude Fable 5 model.

Coinbase CEO Brian Armstrong said the exchange runs models including GLM-5.2 and Kimi in production. According to him, the change cut Coinbase’s overall AI spending nearly in half even as token consumption increased.

Self-hosting is not free. It transfers the cost of GPUs, electricity, maintenance, networking and model operations to the business running the system. The source report describes it as generally most economical for organizations with substantial, sustained AI usage.

How Businesses May Respond to a Ban

For businesses that use DeepSeek through its website or app, a VPN could bypass a U.S. block. Limited app availability and payment restrictions would remain in place. Enterprise enforcement presents a larger challenge because model weights can be downloaded, copied and run offline.

Public repositories such as Hugging Face and independent torrents mirror these files, making complete recall difficult after release. A company can operate a model inside a private, air-gapped data center, limiting regulators’ ability to monitor it. Fine-tuning, quantization and distillation can also combine a Chinese base model with domestic corporate data, making its origin harder to identify.

Businesses may also respond to government pressure rather than a formal prohibition. Government sources cited by Axios said procurement rules, Entity List threats and public pressure campaigns could encourage U.S. firms to stop using Chinese models. The administration will also push to highlight potential backdoors and lack of security with Chinese models, and the governance issue that brings.

Consider how model choice, hosting arrangements, vendor claims and evidence of security controls may affect how customers identify your business as a credible provider.

What to Do Now

  1. Review which AI models operate inside your company, including models installed on private servers rather than accessed through a hosted API.
  2. Document the source, hosting location and operating responsibility for each model, along with the data sent to it.
  3. Compare self-hosting costs with API pricing and include GPUs, electricity, maintenance, networking and staff time in that calculation.
  4. Prepare an alternative for any model that could face procurement restrictions, especially if your business works with federal agencies or contractors.
  5. Make model provenance and security controls visible in your business information and AI-facing content so customers can verify what they are using.

What Critics Say About the Proposal

Former White House adviser Sriram Krishnan and David Sacks, an outside White House AI adviser, oppose the proposed restrictions. According to the source report, they argue that a ban would harm innovation and strengthen the market position of OpenAI and Anthropic, the two leading U.S. AI labs.

Sacks wrote on X: “We are at a critical inflection point in AI policy. The leading closed labs, already a duopoly in terms of AI model revenue, want the government to eliminate their open-source competition.”

The source report says OpenAI and Anthropic may have an interest in the push for restrictions. The broader context includes existing U.S.-China trade tensions. Washington has placed and later eased export restrictions on critical computing hardware, while the Trump administration has stated its intention for the United States to dominate the AI race.

FAQ

What Triggered the Renewed Push to Restrict Chinese AI Models?

Moonshot AI’s release of Kimi K3 prompted the Trump administration to revive earlier efforts to restrict Chinese AI systems in the United States, according to an Axios report dated July 20. Cybersecurity concerns were the stated reason.

Why Are U.S. Companies Adopting Chinese Open-Weight AI Models?

Models such as DeepSeek and Kimi K3 can run on private infrastructure, helping businesses keep data in-house and reduce inference costs. DeepSeek-V4-Pro charges $0.87 per million output tokens, compared with $50 for Anthropic’s Claude Fable 5. Coinbase CEO Brian Armstrong said the exchange’s use of GLM-5.2 and Kimi cut overall AI spending nearly in half.

Can a U.S. Ban on Open-Weight Chinese AI Models Be Enforced?

Enforcement would be difficult because model weights are downloadable files mirrored across repositories such as Hugging Face and can run offline in air-gapped data centers. Companies may also fine-tune, quantize or distill models, making their origin harder to determine. The reported strategy includes procurement rules, Entity List threats and public pressure to encourage companies to stop using the models.

Related coverage

This article reports on publicly available information from the sources listed below. We have no reason to believe it is inaccurate, but we have not independently verified every claim, and details may change as the story develops.

Sources


See our editorial disclaimer for how our articles are produced.

🤖
Is your business visible to AI assistants?

Run a free scan to see your AI Visibility Score, SEO rating, and local citation accuracy.

Check Your Score →