Chinese tech giant Alibaba has made a strategic decision that will take effect on July 10: the use of Anthropic's AI tools, including the popular Claude Code, will be completely banned for all company employees. Instead, developers are instructed to switch to an internal assistant — Qoder.
This move is not just a technical policy update. It is a mirror reflection of the deep geopolitical and technological confrontation between the US and China, which is now entering a phase of active action.
Reason for the ban: vulnerabilities and espionage
The decision was made after Alibaba engineers identified potentially dangerous behavior in Claude Code. The tool, it turned out, collected data on user time zones and proxies, and, according to some reports, inserted hidden tags into requests sent to Anthropic's servers. This practice was deemed a "backdoor" risk — a hidden channel for leaking corporate information.
It is important to note that this incident occurred against the backdrop of escalating tensions between the companies. Earlier, in June, Anthropic sent an official letter to the US Senate Banking Committee, accusing Alibaba of attempting to illegally distill (copy the capabilities of) its AI model. Anthropic called this the "largest distillation attempt" in its practice. Distillation, as a reminder, is a method where the results of one powerful model are used to train a cheaper competing model, allowing development costs to be bypassed.
Part of the bigger picture: the "Qinglang" campaign
The ban on Claude Code is just one element of a large-scale state campaign to tighten control over AI in China, known as "Qinglang." As part of the first phase of this initiative, the Chinese internet regulator has already removed over 14,000 AI products, blocked more than 26,000 accounts, and removed millions of web pages and files that do not meet the new requirements. Alibaba, like other major players, is forced to adapt its services to the tightening rules.
Thus, we are witnessing a classic dilemma of the era of technological nationalism: on one hand, Alibaba loses access to one of the best development tools from the US, which could slow down the pace of software creation. On the other hand, Anthropic loses the opportunity to work with a massive Chinese technology platform. From my point of view, this move is less about data security and more about demonstrating loyalty to state policy and accelerating the transition to proprietary, fully controlled technology stacks. In the long term, this will lead to an even greater division of the global AI market into two isolated camps.