Skip to content
AI.info

The Pulse

Federal Register Removes Alibaba’s Qwen Search Tool

The National Archives removed an Alibaba Qwen-powered search option from the Federal Register website on September 16, 2026. The withdrawal followed public attention and came days after the FBI accused Alibaba of copying Anthropic technolog

Federal Register Removes Alibaba’s Qwen Search Tool

AI.info Team ·

The National Archives removed an Alibaba artificial-intelligence search option from the Federal Register website on Wednesday, September 16, after users drew attention to the tool’s use of China’s Qwen model. The withdrawal came days after the FBI accused Alibaba of copying technology from U.S. rival Anthropic, creating a direct conflict between the government’s public-facing software and Washington’s warnings about Chinese AI.

Reuters reported on September 17 that the Federal Register had offered Qwen-powered search alongside other options for browsing proposed federal regulations and related public comments. Screenshots and an archived version of the site’s source code reviewed by Reuters showed that the Qwen option disappeared on Wednesday, around the time social-media posts began circulating about it.

The National Archives, which operates the Federal Register, has not publicly explained when the Qwen feature was introduced or why it was removed. The agency did not respond to a request for comment from Ars Technica; the White House and FBI also did not comment to that outlet.

A Chinese model inside a federal search page

The Federal Register is the U.S. government’s official journal for proposed rules, final rules, notices and presidential documents. Its search system is designed to help the public locate those records, including comments submitted during federal rulemaking.

Reuters identified Alibaba’s Qwen model as one of the search choices available on the site. Other accounts identified the model as Qwen3 0.6B, a relatively small open-weight system used for document retrieval rather than general-purpose conversation or complex reasoning.

That distinction matters to the security assessment. Georgetown University law professor Anupam Chander told Reuters that the material being searched was already public, meaning the system was not handling classified or otherwise restricted government information. The risk would also depend on whether queries and documents were processed on Alibaba-controlled systems or on infrastructure operated by the U.S. government.

The timing collided with an FBI accusation

The removal followed a week of heightened scrutiny around Alibaba’s AI operations. The FBI accused Alibaba and other Chinese companies of conducting what it described as “industrial-scale distillation,” an allegation that the firms copied capabilities from leading U.S. models to reduce development costs and time.

Reuters’ report did not establish that the Qwen search tool had created a security breach or transmitted sensitive federal information. It did establish that the model had been available on a U.S. government website while federal officials were warning agencies and companies about dependence on Chinese AI technology.

Rep. John Moolenaar, the Michigan Republican who chairs the House China Committee, told Reuters that “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.”

Experts see limited immediate exposure

Technical experts interviewed by Reuters said the particular deployment appeared unlikely to create substantial immediate security risks. A small retrieval model running locally can allow an agency to keep queries and documents within its own infrastructure, rather than sending them to an outside provider.

Qwen’s open-weight licensing also means an organization can download and operate the model without relying on Alibaba’s hosted services. That architecture can give a government office more control over data, although the model’s origin can still raise procurement, supply-chain and national-security questions.

The dispute therefore concerns two separate issues: what data the Federal Register search system handled, and whether a federal agency should use software developed by a company that U.S. officials accuse of appropriating American AI technology. Public documents and comments may present little direct confidentiality risk, but the choice of model can still signal an agency’s tolerance for foreign technology dependencies.

No explanation accompanied the rollback

The Federal Register’s Qwen option vanished without a public announcement, changelog or stated policy decision. Reuters could not determine when the feature went live, who approved it or who ordered its removal.

That leaves open several explanations, including a routine software change, a technical test that ended, or a response to the attention generated by the FBI’s allegations and social-media posts. The available evidence confirms the deployment and its removal, but not the motive behind either decision.

For now, the concrete outcome is limited: a Chinese-developed model that briefly helped users search U.S. regulatory records is no longer listed on the Federal Register website. The National Archives has returned the public-facing service to its remaining search methods without explaining the change.

Source

Reuters

Explore

More articles