Government & Policy | 3 min read

Trump Administration Weighs Rule Forcing U.S. Firms to Disclose Chinese AI Model Use

The Trump administration is considering requiring U.S. companies to disclose publicly when their products are built on Chinese open-weight AI models like DeepSeek, Kimi, or Qwen.

Hector Herrera
Hector Herrera
A government building interior related to Trump Administration Weighs Rule Forcing U.S. Firms to Discl from an unusual angle or perspective
Why this matters The Trump administration is considering requiring U.S. companies to disclose publicly when their products are built on Chinese open-weight AI models like DeepSeek, Kimi, or Qwen.

The Trump administration is considering a rule that would require U.S. companies to publicly disclose when their products are built on Chinese open-weight AI models — a measure designed to reduce American commercial dependence on Chinese AI without imposing outright bans. If enacted, the rule would force companies relying on models like DeepSeek, Kimi, or Qwen to say so on their websites.

The move, reported by Reuters, reflects a broader effort to map — and eventually shrink — the footprint of Chinese AI in U.S. commercial infrastructure, while stopping short of the kind of blanket restrictions that faced legal challenges with TikTok.

The Proposal

The administration is weighing a disclosure requirement, not a ban. The proposed rule would compel U.S. businesses to state on their websites when they deploy Chinese open-weight AI models as a foundation for their products or services. Open-weight models are AI systems whose underlying parameters are publicly released — meaning any company can download, modify, and deploy them without licensing fees or ongoing access to the original developer.

The rule is at an early consideration stage and has not moved to formal rulemaking. No agency has been publicly designated to enforce it, and no draft regulation has been published for comment.

Why It Matters

The proposal puts a number on a problem that the AI policy community has been discussing in the abstract: how deeply embedded are Chinese AI models in U.S. startup infrastructure?

According to an Andreessen Horowitz partner cited in the Reuters report, approximately 80% of U.S. startups currently build on Chinese open-weight models. That figure, if accurate, describes a significant supply-chain dependency — one that policymakers in both parties have flagged as a national security concern, even as engineers and founders have embraced the models for their low cost and competitive performance.

DeepSeek, the Chinese AI lab behind several high-performing open-weight models, became a flashpoint earlier in 2026 when its R1 reasoning model matched or outperformed U.S. counterparts at a fraction of the training cost. Kimi (from Moonshot AI) and Qwen (from Alibaba) have similarly attracted enterprise adoption. All three are open-weight, free to use, and hosted by developers in China.

The Policy Logic

A disclosure-only rule threads a difficult needle:

  • It avoids a ban. Outright prohibition of Chinese AI models would face immediate legal challenges on First Amendment and due process grounds, and would force rapid, expensive migrations for the many startups already embedded in Chinese model ecosystems.
  • It creates visibility. Mandatory disclosure gives regulators, customers, and investors a data layer they currently lack — a public record of which products depend on Chinese AI infrastructure.
  • It applies market pressure. Companies that prefer to avoid the reputational or regulatory risk of a disclosed Chinese AI dependency may voluntarily migrate to U.S. alternatives before any further rules take effect.

Critics of the approach will argue that disclosure without enforcement is regulatory theater — that many startups will comply minimally, bury the disclosure in terms of service, and face no real consequence. Supporters will counter that disclosure is a first step, with more restrictive measures available if transparency alone proves insufficient.

What to Watch

The administration's next move is to determine whether the rule advances to formal agency rulemaking, which would open a public comment period and trigger a longer regulatory timeline. Watch for signals from the Commerce Department or the Office of the U.S. Trade Representative, both of which have jurisdiction over technology and trade policy.

For U.S. AI companies and startups, the practical question is whether this proposal is a warning shot ahead of stricter restrictions — or a ceiling. The fact that it stops at disclosure suggests the administration is calibrating for the political and legal difficulties of a ban, but the 80% dependency figure, if confirmed, will likely drive further action.

Key Takeaways

  • ✓ how deeply embedded are Chinese AI models in U.S. startup infrastructure?
  • ✓ It creates visibility.
  • ✓ It applies market pressure.

Did this help you understand AI better?

Your feedback helps us write more useful content.

Hector Herrera

Written by

Hector Herrera

Hector Herrera is an AI systems architect in Houston and founder of Hex AI Systems. He designs and runs AI systems in production and writes daily about how AI is reshaping business, government and everyday life. 20+ years building for the web. Houston, TX.

More from Hector →

Get tomorrow's AI briefing

Join readers who start their day with NexChron. Free, daily, no spam.

More from NexChron