{"slug": "a-us-government-website-ran-alibaba-s-qwen-days-after-the-fbi-accused-it-of-s", "title": "A US Government Website Ran Alibaba's Qwen Days After the FBI Accused It of Stealing Anthropic's Work", "summary": "The Federal Register, run by the National Archives, removed an AI search tool built on Alibaba's Qwen model on Wednesday, September 17, 2026, the same day social media posts flagged it, according to Reuters, which reviewed screenshots and an archived copy of the site's source code. The takedown came days after the FBI publicly accused Alibaba of \"malicious\" copying of American AI technology on an \"industrial scale,\" and after Anthropic told lawmakers in June that operators tied to Alibaba and its Qwen research unit ran a distillation campaign against its models between April 22 and June 5, 2026, using 25,000 fake accounts to harvest 28.8 million exchanges from Claude. Senator Mark Warner, vice chairman of the Senate Intelligence Committee, said the unanswered question is whether US data left the government's security boundary and was processed on servers Alibaba controls; it remains unclear when the tool was deployed, who approved it, or where queries were routed.", "body_md": "*The Federal Register, the site where the public searches proposed federal rules, quietly ran an AI search tool built on Alibaba's Qwen model, then pulled it the same day social media posts called it out.*\n\nYou'd think the agency publishing America's official rulebook would be the last place to find a Chinese AI model the FBI just called a thief. But that's exactly what happened. The Federal Register, run by the National Archives, had a search feature powered by Alibaba's Qwen model live on its site as of Wednesday, September 17, 2026, sitting right alongside its other search options. According to Reuters, which reviewed screenshots and an archived copy of the site's source code, the tool came down around the same time posts on social media started pointing it out.\n\nThe timing could not be worse for Washington's China hawks. Days earlier, the FBI publicly accused Alibaba of what US officials called \"malicious\" copying of American AI technology on an \"industrial scale.\" Anthropic told lawmakers in June that operators tied to Alibaba and its Qwen research unit ran a large-scale \"distillation\" campaign against its models between April 22 and June 5, 2026, essentially using Anthropic's own outputs to train a rival system. The Chinese embassy in Washington called the copying allegations \"unfounded.\"\n\nSo a federal agency ran the accused software on its own infrastructure while another arm of the federal government was branding it a security threat. That's not a coincidence anyone planned. It's a coordination failure.\n\n## How Sensitive Was This, Really?\n\nNot very, in this specific case. The Federal Register's content, proposed regulations open for public comment, is already public by design, so there was no classified or sensitive government data sitting behind the Qwen-powered search. Democratic Senator Mark Warner of Virginia, vice chairman of the Senate Intelligence Committee, put the real question plainly in a statement: whether US data left the government's security boundary and got processed on servers Alibaba controls. That's the detail nobody has nailed down yet. It's unclear when the tool was first deployed, who approved it, or where the queries were actually routed once a member of the public typed something in.\n\n[China Publicly Rejects Anthropic's Claim That Alibaba Stole Claude's Data](https://startupfortune.com/china-publicly-rejects-anthropics-claim-that-alibaba-stole-claudes-data/)\n\nChina's assistant foreign minister Liu Bin dismissed US distillation accusations as \"misguided and counterproductive\" at the World AI Conference in Shanghai. He never addressed the specific numbers behind Anthropic's claim that Alibaba's Qwen lab used 25,000 fake accounts to harvest 28.8 million exchanges from Claude. - [China's response to Claude distillation claims](https://startupfortune.com/china-publicly-rejects-anthropics-claim-that-alibaba-stole-claudes-data/) - [Alibaba stealing Anthropic AI model data](https://startupfortune.com/china-publicly-rejects-anthropics-claim-that-alibaba-stole-claudes-data/)\n\nHere's the thing. Low sensitivity does not mean low stakes. A federal website is federal infrastructure, and the whole point of restricting Chinese AI models from government systems is to avoid exactly this kind of quiet, unnoticed adoption. Nobody flagged it internally. It took outside observers on social media to force the takedown.\n\nThis is what procurement blind spots look like in practice. Federal agencies buy or integrate software constantly, often through contractors, plugins, or vendor add-ons that nobody at the top explicitly signs off on. A search widget powered by a foreign model can get baked into a government site without ever crossing the desk of anyone thinking about export controls or AI provenance. The Federal Register episode is a small, low-stakes version of a much bigger problem: nobody appears to be running a consistent check on which AI models sit inside federal infrastructure.\n\n## The Worst Possible Timing\n\nThe politics land at the worst possible moment. President Trump and President Xi Jinping are set to meet in Washington on September 24, the first Chinese state visit to the US since 2015, with AI competition, trade, and Taiwan all on the agenda, according to CNBC's reporting on the summit. Walking into that meeting, US officials want the upper hand: a clean case that Chinese firms steal American AI work and deserve restriction. Instead, reporters get to ask why a US government website was running the accused firm's model days before the FBI's accusation became public.\n\nFrankly, this undercuts the message before it's even delivered. You can't credibly tell Beijing that Alibaba's AI is untrustworthy. Not while your own National Archives is quietly serving search results through it. Critics of the administration's China AI policy now have a concrete, screenshot-able example instead of an abstract argument about supply chain risk.\n\nAnthropic, for its part, has leaned into its role as the company sounding the alarm on Chinese model theft, a stance that doubles as competitive positioning against Qwen and other rivals gaining ground on cost and capability. Whether that argument gains or loses credibility now depends less on Anthropic's evidence and more on whether the federal government can show it actually knows what AI software is running on its own sites. Right now, the answer looks like no.\n\n**Also read:** [CrowdStrike and Palo Alto Soared on AI Fears, Then Bernstein Called the Top](https://startupfortune.com/crowdstrike-and-palo-alto-soared-on-ai-fears-then-bernstein-called-the-top/) • [OpenAI's Own Numbers Show It Could Run Out of Cash by 2028](https://startupfortune.com/openais-own-numbers-show-it-could-run-out-of-cash-by-2028/) • [An AI Chatbot Nearly Triggered a US Strike on a Chinese Ship](https://startupfortune.com/an-ai-chatbot-nearly-triggered-a-us-strike-on-a-chinese-ship/)\n\n[Anthropic Says China Was Never Authorized to Use Claude Code Anyway](https://startupfortune.com/anthropic-says-china-was-never-authorized-to-use-claude-code-anyway/)\n\nChina's Ministry of Industry and Information Technology warned that Anthropic's Claude Code contains a hidden monitoring mechanism sending user data to remote servers. Anthropic confirmed the code exists but said the affected users in China were never authorized to use Claude Code in the first place. - [claude code china unauthorized access](https://startupfortune.com/anthropic-says-china-was-never-authorized-to-use-claude-code-anyway/) - [anthropic tracking mechanism developer tools](https://startupfortune.com/anthropic-says-china-was-never-authorized-to-use-claude-code-anyway/)\n\n*This article is posted in [AI News](https://startupfortune.com/category/ai/), check it out for more related stories.*\n\n## Join the discussion\n\n[Open in the community →](https://startupfortune.com/community/)\n\nAlmost there. Sign in and your reply posts straight away.", "url": "https://wpnews.pro/news/a-us-government-website-ran-alibaba-s-qwen-days-after-the-fbi-accused-it-of-s", "canonical_source": "https://startupfortune.com/a-us-government-website-ran-alibabas-qwen-days-after-the-fbi-accused-it-of-stealing-anthropics-work/", "published_at": "2026-09-19 04:54:07+00:00", "updated_at": "2026-09-19 05:22:55.212095+00:00", "lang": "en", "topics": ["ai-policy", "ai-safety", "large-language-models", "ai-search"], "entities": ["Federal Register", "National Archives", "Alibaba", "Qwen", "FBI", "Anthropic", "Mark Warner", "Liu Bin"], "alternates": {"html": "https://wpnews.pro/news/a-us-government-website-ran-alibaba-s-qwen-days-after-the-fbi-accused-it-of-s", "markdown": "https://wpnews.pro/news/a-us-government-website-ran-alibaba-s-qwen-days-after-the-fbi-accused-it-of-s.md", "text": "https://wpnews.pro/news/a-us-government-website-ran-alibaba-s-qwen-days-after-the-fbi-accused-it-of-s.txt", "jsonld": "https://wpnews.pro/news/a-us-government-website-ran-alibaba-s-qwen-days-after-the-fbi-accused-it-of-s.jsonld"}}