The Trump administration is considering requiring US companies to publicly disclose their use of Chinese AI models, three people told Reuters on Wednesday. The plan would apply to companies such as Airbnb, which Reuters says uses Alibaba’s Qwen to power its AI customer service agent.
Our number: on OpenRouter’s leaderboard for this month, with usage data through 17 November, 8 of the 10 most-used models come from Chinese developers, and they account for about 87% of the top 10’s tokens, by our count and arithmetic.
Business Pill · Embedded Dependency
An evergreen Business Engineer explainer on embedded dependency, what is inside what you buy. It teaches the general idea only and does not discuss this proposal.
The key insight: Reuters reports the administration is weighing a disclosure requirement, not a ban, for US companies using Chinese AI models; on OpenRouter this month 8 of the 10 most-used models come from Chinese developers, about 87% of the top 10’s tokens by our arithmetic.
What Reuters Reported
According to Reuters, the Treasury Department and the White House Office of Science and Technology Policy have discussed the plan. Under it, companies like Airbnb would be required to post that information on their websites.
Reuters reported that the plan would aim to discourage reliance on Chinese AI without imposing new restrictions on US firms. It said the proposal shows officials looking at ways to discourage use of Chinese AI through a “name and shame” campaign while sticking to a broader anti-regulation stance on AI.
Reuters also wrote: “It was not immediately clear if the idea was gaining traction within the administration or if it would be enshrined in a formal rulemaking process if it does move forward”. The White House, the Treasury Department, Alibaba and the Chinese Embassy in Washington did not immediately respond to Reuters’ requests for comment, and Airbnb declined to comment.

Who Is Already Being Asked
Republican Representatives Andrew Garbarino and John Moolenaar, who chair the House Homeland Security Committee and the House select committee on China, launched an investigation into American companies’ use of Chinese AI this year, Reuters reported. They sent letters to Airbnb, to Anysphere, the maker of Cursor, whose Composer 2 model Reuters says was built on a version of Kimi, and to DoorDash, whose co-founder said the company routes some tasks in an internal system through Kimi.
Reuters said Chinese-made open-weight models are gaining ground with US companies because they are cheaper than proprietary tools from OpenAI and Anthropic and perform well enough. It reported that China hawks fear Beijing is using models such as Kimi, DeepSeek or Qwen to make US companies dependent on Chinese AI.
What the Usage Data Shows
Reuters cited OpenRouter data showing Chinese AI models were the three most popular by token usage in October. On OpenRouter’s own leaderboard for this month, DeepSeek V4.1 Flash 1017 leads with 215T tokens. The top 10 includes two DeepSeek models, two GLM models from Z.ai, three Tencent Hy4 models and StepFun’s Step 5 Preview.
The two US models in the top 10 are OpenAI’s GPT-6 Luna with 35.6T tokens and Anthropic’s Claude Opus 5.5 with 31.4T. By our arithmetic, the eight Chinese models processed about 463.2T tokens and the two US models 67.0T, so the Chinese models account for about 87% of the top 10’s 530.2T. We covered DeepSeek’s lead earlier here.
The proposal adds to a week of China measures: the Commerce Department’s rule on remote access to advanced chips and the Trump-Xi meeting in Shenzhen.
The Structural Read
As we read the Reuters report, the tool under discussion is publicity rather than restriction: companies would post their use of Chinese models on their websites, and Reuters describes the aim as discouraging reliance without new restrictions.
As we read the OpenRouter figures, the usage the proposal targets is concentrated in a few developers: DeepSeek, Z.ai, Tencent and StepFun account for 8 of the 10 most-used models this month.
Reuters, 18 November 2026
“It was not immediately clear if the idea was gaining traction within the administration or if it would be enshrined in a formal rulemaking process if it does move forward.”
Three Implications
FOR COMPANIES USING CHINESE MODELS Reuters reports the plan would apply to companies such as Airbnb, which uses Qwen for its AI customer service agent, and would require posting that information on their websites.
FOR CURSOR AND DOORDASH Reuters reports House committee chairs sent letters to Anysphere, whose Composer 2 model was built on a version of Kimi, and to DoorDash, which routes some internal tasks through Kimi.
FOR MODEL DEVELOPERS OpenRouter’s month leaderboard lists DeepSeek V4.1 Flash 1017 first with 215T tokens, ahead of GPT-6 Luna at 35.6T and Claude Opus 5.5 at 31.4T.
The Business Engineer Lens
This story maps onto the Business Engineer framework The Open vs Closed Meta-Framework.
The framework starts from this premise: “Every technology stack has layers. Some are scarce, some are abundant.”
As we read the OpenRouter figures, Chinese open-weight models are competing in the layer the framework calls abundant, on price, which is the reliance the reported disclosure plan is meant to make visible. The framework predates this proposal and does not discuss it.
What Is Not Established
Reuters described the proposal as under consideration. It was not immediately clear whether it is gaining traction or would go through a formal rulemaking, and no official text was available in the sources we read.
OpenRouter’s figures cover usage routed through its own platform, not total AI usage by US companies. The classification of the eight models as Chinese is ours, based on the developers OpenRouter lists. We did not contact the White House, the Treasury, Airbnb or DoorDash.
The Bottom Line
Reuters reports that the administration is weighing a disclosure requirement rather than a ban on Chinese AI models. On OpenRouter this month, 8 of the 10 most-used models come from Chinese developers, about 87% of the top 10’s tokens by our arithmetic.
97,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. The proposal, the Treasury and OSTP discussions, the companies named and the congressional letters are as reported by Reuters, citing three people; Reuters says it was not immediately clear whether the idea is gaining traction. Model rankings and token counts are from OpenRouter’s public leaderboard (This Month view, usage data through 17 November 2026). The count of Chinese models and the 87% share are our arithmetic on those figures. This piece is not investment advice.
Sources: Reuters, 18 Nov 2026 · OpenRouter LLM leaderboard, month view (data through 17 Nov 2026)









