SIAgentNews.com
← All news

U.S. weighs name-and-shame strategy to deter Chinese AI use

The U.S. government is considering a plan to require companies to disclose their use of Chinese AI models to the public as a way to discourage adoption.

· First reported by The Chosun Ilbo on MSN

46%Peak ratio of Chinese model tokens on OpenRouter

The short version

  • The U.S. may require companies to post public disclosures on their websites regarding the use of Chinese AI models.
  • Chinese models ranked first to third in monthly token usage on OpenRouter in October.
  • Over 30% of tokens used by companies on OpenRouter were from Chinese models since February.
  • The plan aims to deter usage through public exposure rather than direct new regulations.
  • The U.S. Treasury and White House Office of Science and Technology Policy are currently discussing the proposal.

Why it matters

This represents a shift in U.S. strategy from blocking China's ability to build AI to preventing American companies from adopting Chinese AI. It could force U.S. firms to choose between cheaper Chinese technology and avoiding potential public backlash.

The name-and-shame strategy

The Donald Trump administration is exploring a new policy to manage the presence of Chinese artificial intelligence within the United States. Instead of creating new, direct regulations, the government is considering a "name-and-shame" strategy to influence corporate behavior. This approach would involve mandating that U.S. companies clearly disclose on their websites whether they utilize Chinese AI models.

The goal of this disclosure requirement is to create a deterrent by exposing the use of Chinese technology to the general public. By making these connections visible, the government hopes to discourage American firms from integrating these specific models into their operations. The U.S. Treasury Department and the White House Office of Science and Technology Policy are currently the primary bodies discussing the proposal.

This move comes as U.S. officials become increasingly hawkish toward Chinese technology. The focus is shifting from strictly controlling China's ability to develop high-end hardware to preventing the spread of Chinese AI software within the domestic market. The administration wants to ensure that American companies are not becoming reliant on Chinese infrastructure for their core services.

The rise of Chinese AI models

Chinese AI models are rapidly gaining traction among U.S. firms primarily because they are highly cost-effective. Data from OpenRouter, a platform that mediates different AI models, shows that Chinese models ranked first to third in monthly token usage last October. A token is a basic unit of data, such as a word or part of a word, that an AI processes to generate a response.

The scale of this adoption is significant, with reports indicating that over 30% of tokens used by companies on OpenRouter were from Chinese models starting in February. This usage ratio reached a peak of 46% during that same period. These figures suggest that many American businesses are already integrating Chinese models into their daily workflows and software products.

Several prominent American companies have already been identified as users of Chinese AI. For example, Airbnb uses Alibaba’s "Qwen" model for customer service tasks. Additionally, the food delivery platform DoorDash disclosed that one of its co-founders used a model called "Kimi K2.6" for parts of its code review work. Even the U.S. Federal Register website was found to have used an Alibaba model in September.

Corporate and political fallout

The push for disclosure follows a period of intense political scrutiny and a specific scandal involving NVIDIA founder Jensen Huang. The scandal, which involved allegations of an affair with a suspected Chinese spy, has reinvigorated the U.S. government's hardline stance on Chinese technology. The tech industry views the current disclosure discussions as a direct backlash resulting from that situation.

In response to the growing scrutiny, some companies have already begun clarifying their AI practices. Airbnb issued a statement explaining that its AI primarily operates on U.S.-based models. They noted that while they use a small number of open-source Chinese models, they only do so through approved U.S. vendors to protect their data and operations.

The U.S. Congress has also taken active steps to investigate these connections. This year, lawmakers sent letters to companies like Airbnb, Anysphere, and DoorDash to probe their specific use of Chinese AI. These inquiries highlight a growing concern that American corporate data and infrastructure may be becoming entangled with Chinese technology providers.

Barriers to implementation

Despite the current discussions, there is significant uncertainty regarding whether this disclosure mandate will actually become law. The U.S. government has struggled to move from proposal to implementation in the past. Reports indicate that four separate attempts to restrict Chinese AI models were scrapped last year after undergoing internal administration reviews.

The Treasury Secretary Scott Bessent recently highlighted a specific security concern regarding these models. He noted that officials have found watermarks of U.S. large language models in many Chinese models, calling the situation unacceptable. He suggested that whether companies should be required to disclose this to their customers remains a primary question for the administration.

The Chinese government has already pushed back against these potential moves. The Chinese Embassy in the U.S. criticized the strategy as an attempt to politicize trade and technology. They argued that such actions will only hinder global AI development and vowed to protect the legitimate rights of their own companies in the international market.

What to watch

  • Whether the Treasury Department and OSTP finalize the disclosure mandate.
  • How other U.S. companies react to the potential for public 'naming and shaming.'
  • If the U.S. moves from disclosure requirements to direct bans on specific Chinese models.

Try it

Check the websites of major U.S. tech companies to see if they have a dedicated section or disclaimer regarding their AI providers and data privacy.

Sources and links

First reported by The Chosun Ilbo on MSN. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

More AI news