- The Trump administration is considering requiring U.S. companies to disclose publicly when they use Chinese artificial intelligence models.
- The proposal is intended to curb reliance on Chinese technology without direct restrictions that could provoke opposition from businesses or Beijing.
- The discussion comes as models from Alibaba, DeepSeek and Moonshot AI gain popularity among U.S. companies.
The Trump administration is considering requiring U.S. companies to publish information about their use of Chinese artificial intelligence models. The measure could expose businesses to public scrutiny as Washington seeks to limit dependence on Chinese developers while avoiding direct restrictions.
The Treasury Department and the White House Office of Science and Technology Policy discussed the initiative, Reuters reported, citing three sources.
It remains unclear whether the proposal has enough support within the administration to advance to formal regulation. Officials are considering it as a possible tool for a “name and shame” campaign, under which companies would publish the disclosures on their own websites.
The approach could allow Washington to discourage businesses from adopting Chinese AI technology without imposing restrictions that could trigger pushback from companies or Beijing. Airbnb could be covered because it uses Qwen, a model developed by China’s Alibaba, to power its customer-service AI agent.
Chinese models gain traction among U.S. companies
U.S. officials are concerned about the spread of open-weight models, whose core components are available for users to download and adapt. Chinese models including Alibaba’s Qwen, DeepSeek and Moonshot AI’s Kimi offer lower costs than closed products from leading U.S. laboratories OpenAI and Anthropic while providing sufficient performance for a range of tasks.
In October 2026, the three most popular models by the number of tokens processed were Chinese, according to AI model-usage platform OpenRouter. Advocates of a tougher U.S. policy toward Beijing say that popularity could deepen U.S. companies’ technological dependence on Chinese developers.
U.S. officials have also raised concerns that American technology could be used to build competing Chinese models. In July, Treasury Secretary Scott Bessent said the administration could impose sanctions on Chinese AI companies if it is proven that they stole U.S. developers’ intellectual property through model distillation.
Model distillation involves training smaller models on the outputs of larger ones to reduce the cost of developing new systems. In April, OSTP Director Michael Kratsios accused China of stealing U.S. AI laboratories’ intellectual property “on an industrial scale.” The State Department subsequently began a campaign to inform foreign partners about such cases.
Congress examines corporate use
The issue has also drawn scrutiny from Congress. This year, Republican Representatives Andrew Garbarino and John Moolenaar, who chair the House Homeland Security Committee and the House Select Committee on China, respectively, opened an investigation into U.S. companies’ use of Chinese AI.
They sent inquiries to Airbnb, Cursor developer Anysphere and DoorDash. Cursor’s Composer 2 model is built on a version of Kimi, while a DoorDash co-founder said the company routes some tasks from its internal system through Kimi.
U.S. companies are also developing open models despite concerns about Chinese technology. Those models include Meta’s Muse Glimmer, Nvidia’s Nemotron and Reflection AI’s Beam.
Source: Incrypted
