The US Federal Register briefly used a Chinese artificial intelligence model to search its public documents, then removed the system two days later. The episode has opened a much bigger debate: can the US government use open-source AI from China when the tool handles public information and runs locally?
The model was Alibaba’s Qwen3 0.6B level system, a small open-weight model built for document retrieval. It appeared on the Federal Register website on September 15, 2026, and the government took it down on September 17 after social media posts and reports drew attention to its use.
A Small Model With a Very Narrow Job
Qwen3 0.6B was not being used as a broad government assistant. The model served solely to retrieve documents from the Federal Register, the US government’s official journal, whose content is already public.
That detail matters because the system was not handling sensitive government information. The model also did not provide complex reasoning, and it did not send government data to Alibaba or any third parties. As an open-source model, Qwen could be downloaded and run locally instead of sending search requests to an outside service.
This setup led experts to conclude that the deployment posed no substantial security risks. The Federal Register’s public content meant the model was searching material already available to anyone, while its small size and limited purpose placed clear boundaries around what it could do.
Still, the model’s origin turned a routine search feature into a national policy flashpoint. Alibaba is a Chinese AI company, and the FBI called the Chinese model “malicious.” The FBI also accused Alibaba of “malicious” copying of US rival Anthropic’s technology, adding a technology competition dispute to the government’s decision to remove Qwen.
Security Concerns Collide With AI Costs
The controversy reaches beyond this single Federal Register feature. Joe Wilkins captured one side of the pressure when he said, “Even the United States government is unable to afford the cost per token charged by American AI companies.” That argument points to a practical challenge: public agencies may want advanced AI tools, but the cost of using models from American companies can shape which systems they choose.
The opposing concern focuses on dependence. US Rep. John Moolenaar, the chair of the House China Committee, said, “no federal government entity should use a Chinese AI model.” He added, “Doing so only makes the federal government more dependent on Chinese AI models, and that is not in the national interest.”
The FBI suggested that Chinese models could create national security concerns by locking the US government into Chinese technology. Experts pushed back on the specific Federal Register deployment, saying the Qwen model’s narrow role and access to public content made substantial risks unlikely.
That split reveals why the incident has drawn attention even though the tool had limited capabilities. One side sees a small, local, open-source search model working on public documents. The other sees any government use of Chinese AI as a step toward dependence, regardless of the model’s size or task.
What Happens After the Removal?
The government’s timeline was short but clear: Qwen was available on September 15, social media posts and reports brought the use into view, and the Federal Register removed it on September 17. The Ars Technica article about the episode was published on September 18, extending the discussion beyond the website itself.
The removal does not erase the practical question behind the controversy. If a small open-weight model can search public records without sending data outside the government, does its country of origin alone make it unacceptable? The answer will shape how federal agencies evaluate open-source AI, especially when public budgets meet the high cost of commercial systems.
The Federal Register experiment also shows how fast an ordinary software choice can become a national technology debate. A model designed only to retrieve documents triggered concerns from US authorities, criticism from a senior lawmaker, and a broader argument over whether local operation can reduce the risks tied to foreign AI.
For now, the Qwen model is gone from the Federal Register website. The larger contest over AI cost, open-source access, security, and US dependence on Chinese technology is still moving forward.
Based on
- US Federal Register Caught Using Chinese AI Model for Document Search — futurism.com
- US government website used Chinese model the FBI called “malicious” – Ars Technica — arstechnica.com
- US government website used Chinese model the FBI called “malicious” | Ars OpenForum — arstechnica.com
- US government website used AI search tool from China that FBI said copied Anthropic | Reuters — reuters.com
