Alibaba fully bans Claude Code! From employee suspension to promoting self-developed Qoder, the geopolitical AI war escalates again.

Alibaba orders a full discontinuation of Claude Code starting July 10, lists Anthropic’s AI programming tool as high-risk software, and requires employees to switch to the group’s self-developed Qoder. This is the latest round of escalation in the US-China AI rivalry, driven by the prior “distillation attack” controversy, the Claude Code spy-code incident, and Alibaba’s series of actions to completely block Claude products.
(Background: Alibaba ordered all staff to uninstall the entire Claude suite; the distillation-attack defendant retaliated by issuing a ban.)
(Background supplement: Claude Code admitted to embedding “spy code” for Chinese users to prevent resale and distillation, and removed it only after it was exposed.)

Table of Contents

Toggle

• The Claude Code spy-code incident
• From distillation attacks to a full shutdown: a timeline of the confrontation
• US-China AI decoupling: from the chip wars to development tools

Alibaba Group, a Chinese e-commerce giant, has ordered that starting July 10, employees are comprehensively prohibited from using Anthropic’s AI programming tool Claude Code, which it classifies as high-risk software, and employees are required to switch to the group’s self-developed software development tool Qoder. The ban covers Alibaba’s employees worldwide. According to reports by multiple foreign outlets including Reuters and Dow Jones, the underlying reason is related to concerns about “backdoor risk.”

The Claude Code spy-code incident

In fact, Anthropic had already banned Chinese enterprises and overseas entities owned by Chinese capital from using its models. However, a recent leak on Reddit has revealed a more sensitive internal detail: Anthropic had implanted a version in Claude Code that could secretly identify Chinese users, sparking a “spyware” controversy.

Anthropic’s Thariq Shihipar responded on the X platform, saying that the mechanism “was an experiment we launched in March, aimed at preventing misuse by unauthorized reseller accounts and protecting the model from distillation attacks (distillation).” Distillation attacks refer to the practice of using the output of other AI models to train one’s own model. In the past, Anthropic had taken Alibaba to court over this issue.

From distillation attacks to a full shutdown: a timeline of the confrontation

Looking back at the conflict between Anthropic and Alibaba, a clearly escalating line of confrontation can be seen. In April this year, Anthropic publicly accused Alibaba of carrying out the “largest clone attack in history,” using at least 28.80 million API calls to conduct distillation attacks on the Claude model and steal its capabilities. Soon after, Alibaba ordered all employees to remove the entire Claude product suite, and the relationship between the two sides officially broke down. Immediately thereafter, Claude Code was revealed to have embedded a Chinese user-identification mechanism, triggering the “spy-code” controversy. Although Anthropic argued it was an experiment to prevent misuse, it had already driven trust from China’s tech industry down to freezing point.

Now, Alibaba has formally added Claude Code to its list of high-risk software and requires employees to switch to the self-developed Qoder as an alternative, signaling that this competition over AI tools has moved from “passive blocking” to “active replacement.”

US-China AI decoupling: from the chip wars to development tools

This conflict is not only a commercial dispute between companies; it also reflects the deeper trend of US-China AI decoupling. From NVIDIA’s high-end chip export controls, to OpenAI and Anthropic blocking access to Chinese IP, and now Alibaba’s comprehensive discontinuation of US AI development tools, the split in the AI supply chain between the two sides is expanding from the hardware level to the software tool level.

Alibaba’s strong promotion of its self-developed Qoder is a microcosm of China’s tech industry strategy of “independent, controllable AI.” When US AI companies impose technological restrictions on China, China’s tech giants have no choice but to accelerate the development of domestic alternative solutions, even if those alternatives still lag in functionality in the short term.

For developers in Taiwan, the lesson from this conflict is that AI development tools are becoming a new battleground for geopolitics. When choosing tools, one must consider not only technical capability but also supply-chain risk. Currently, Anthropic’s Claude series products can still be used normally in Taiwan, but the continued escalation of US-China technology confrontation may further affect the global ecosystem layout of AI development tools.