Alibaba Bars Its Staff From Using Claude Code

The rift between Anthropic and Alibaba has widened into an outright ban. From July 10, Alibaba will stop its employees from using Claude Code, Anthropic's command-line coding tool, having placed it on the company's high-risk software list. TechCrunch and the South China Morning Post reported the move, which sends staff to Alibaba's own coding platform, Qoder, instead.

What set it off

The trigger was a piece of code. Security researchers posting on Reddit and GitHub found that a version of Claude Code had been built to examine a user's local environment, reading details such as timezone settings and proxy configuration, and to quietly embed identifying markers in the data it sent back to Anthropic. To Alibaba, that looked like a tool phoning home with fingerprints of the machines running it.

Anthropic's account is narrower. Thariq Shihipar, who works on the tool, wrote on X that the behaviour came from an experiment launched in March that was meant to prevent account abuse from unauthorised resellers and to protect against distillation. He added:

The team has landed stronger mitigations since then and we've actually been meaning to take this down for a while.

In other words, Anthropic frames the code as an anti-abuse measure it had already outgrown, while Alibaba treats it as a security risk serious enough to bar from its network.

The backstory that made it personal

This did not come out of nowhere. In late June, Anthropic accused Alibaba of running what it called the largest known distillation attack on its models. In a June 10 letter to two US senators, the company said entities affiliated with Alibaba had used roughly 25,000 fraudulent accounts to run nearly 28.8 million exchanges with Claude over about six weeks, then used the outputs to train competing systems. Days later, Anthropic moved to close the offshore routes Chinese firms had used to reach Claude.

The account-inspection feature and the distillation accusation are two sides of the same fight. Anthropic built controls to stop what it saw as scraping and abuse. Alibaba, on the receiving end of those controls, has now decided the tool itself is untrustworthy.

A cleaner split every week

The practical result is another strand of the US and China AI markets pulling apart. Alibaba's engineers move to a home-grown tool, Anthropic loses a large pool of Chinese users it was already trying to wall off, and each company can point to the other as the reason. For a working developer in Hangzhou, the effect is simple enough: come July 10, Claude Code stops being an option.