On Wednesday, the US government website removed a Chinese AI search tool, Qwen, after social media users spotted the contradiction. The National Archives had been using Alibaba’s Qwen model, even as the FBI warned of industrial-scale distillation by Chinese companies. The FBI named Alibaba among six firms allegedly cutting costs and development time in the global AI race. It’s unclear when the Qwen model was added to the Federal Register website, and the National Archives have not commented on its removal.

The White House and FBI have also remained silent. Anupam Chander, a Georgetown University Law Professor, noted the “insane” disconnect for a US agency to use an Alibaba model while the FBI encourages American models. Whether this use poses a security risk is uncertain, depending on how the model was trained and if any US data was processed by Alibaba-controlled systems.

On X, a user with the handle “tleilax___” shared a screenshot showing the Qwen model on the government website on September 15. Reuters confirmed that the service was available for at least a day before its removal.

While the use of Qwen may not pose immediate security risks, the incident highlights the complex interplay between technology, geopolitics, and national security. It’s a reminder that in the AI race, every byte can be a diplomatic fowl.