The workplace ban, starting July 10, lands weeks after Anthropic accused operators linked to Alibaba’s Qwen lab of running the largest known distillation campaign against Claude.
Alibaba will bar its employees from using Anthropic’s Claude Code inside workplace environments from July 10, according to a person familiar with the matter cited by Reuters.
The stated reason is an alleged backdoor built into the coding tool, though Alibaba has not confirmed the move publicly and did not immediately respond to a request for comment.
The ban was first reported by the Chinese financial outlet Yicai before Reuters corroborated it through its own source. It arrives at an already tense moment for Anthropic and Alibaba, whose AI units have spent the past two months accusing each other of bad behaviour, first over alleged model theft and now over an alleged spying mechanism baked into Claude’s own tooling.
The 💜 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!
Claude Code is Anthropic’s command-line coding agent, used by developers to write and debug software from a terminal rather than a chat window. It has become one of the company’s fastest-growing enterprise products, which is part of why a workplace-wide ban at a company the size of Alibaba is notable.
The alleged backdoor traces back to a Reddit post published on June 30 by a user identified as LegitMichel777, who said they had reverse-engineered Claude Code while restoring a disabled remote-control feature.
According to a technical write-up shared alongside the post and later summarised by outlets including CyberSecurity News and Tech Times, the coding assistant had quietly checked, since version 2.1.91 released on April 2, whether a user’s proxy configuration or system timezone matched entries on two hidden lists.
One list allegedly named Chinese corporate networks, cloud regions and AI labs, including Alibaba, Baidu, ByteDance and Moonshot AI. If a match was found, the tool reportedly altered the date format and swapped a punctuation character in its own system prompt to encode the detection, rather than sending an overt telemetry signal.
Anthropic has not issued a formal public statement on the allegation. A member of its Claude Code team, Thariq, is reported to have responded on social media that the mechanism was meant to curb account reselling and model distillation, and that it would be stripped out in the next release, a fix The Register and others reported was already underway by July 1.
That timeline means the mechanism was reportedly still live for roughly three months before its removal.
None of this happened in isolation. In a letter dated June 10 to US senators, Anthropic accused operators connected to Alibaba’s Qwen AI lab of running nearly 25,000 fraudulent accounts to extract Claude’s software engineering and reasoning capabilities, generating more than 28.8 million exchanges between April 22 and June 5.
We have reported at the time that the campaign exceeded the combined scale of three earlier distillation efforts Anthropic had already flagged to Washington, including ones it attributed to DeepSeek, Moonshot and MiniMax. Alibaba has not commented publicly on that accusation either.
The dispute sits alongside a broader pattern of restrictions tech companies have placed on coding agents amid distillation fears, and Anthropic’s own tightening of access for Chinese users through tools like Claude Opus and Fable model curbs.
Whether the alleged backdoor was a targeted espionage tool or a blunt anti-fraud filter that swept up ordinary Chinese-based developers remains contested, and no independent security firm has yet published a full audit of the claim.
Alibaba’s ban, if it proceeds as described on July 10, would make it one of the first major companies to formally restrict Claude Code specifically over the alleged mechanism rather than over competitive or cost concerns. Chinese developers who rely on proxy routing to reach the tool at all would be among those most exposed if the detection worked as the researcher described.
Reuters said its report was based on a single source and that Alibaba had not responded by the time of publication. Anthropic was not quoted directly in the Reuters report either, leaving both companies’ full positions on the record still unclear as the July 10 deadline approaches.
