Alibaba has ordered employees to stop using Anthropic’s Claude Code, citing security concerns and potential data risks, according to reports from Reuters and Morningstar. The ban, effective July 10, follows allegations that the AI tool was used to identify users in restricted regions. Alibaba is now directing staff toward its proprietary coding assistant, Qoder.
Why Alibaba is restricting Claude Code
Alibaba has officially classified Anthropic’s Claude Code as high-risk software, prompting an internal directive for staff to abandon the tool. This decision follows growing scrutiny over how AI models handle user data and identify geographic locations. According to reports cited by Reuters, the move is a direct response to perceived security vulnerabilities that could compromise the company’s internal development environment.
Distillation, a term often cited in AI security discussions, refers to the process of training a smaller AI model using the outputs of a larger, more powerful model. Anthropic has expressed concerns that its technology is being used for this purpose without authorization.
The role of geofencing and user identification
Anthropic maintains a strict policy prohibiting Chinese companies and their foreign subsidiaries from accessing its AI models. To enforce these restrictions, the company has actively sought to close technical loopholes. A recent discussion on Reddit alleged that a version of Claude Code contained hidden mechanisms to identify Chinese users. Thariq Shihipar, representing Anthropic, addressed these claims on X (formerly Twitter), describing the mechanism as “an experiment we launched in March that was meant to prevent account abuse from unauthorized resellers and protect against distillation.”
The shift toward proprietary AI tools
As international tech firms tighten access to their models, companies like Alibaba are accelerating the deployment of internal alternatives. Alibaba is instructing its workforce to transition to Qoder, its own in-house coding tool. This shift highlights a broader industry trend where major corporations are moving away from third-party AI dependencies to maintain control over proprietary source code and data privacy. By keeping development within a closed, internal ecosystem, firms aim to mitigate the risks of “backdoor” access or unauthorized data harvesting that regulators and security teams now monitor closely.
How security policies impact global AI adoption
The tension between AI developers and global corporations is shaping how software is built. While Anthropic continues to refine its mitigations to prevent unauthorized use, the fallout has created a fragmented landscape for developers. Shihipar noted that the team has implemented “stronger mitigations” since the March experiment and had been planning to remove the experimental code. Despite these clarifications, the incident underscores the difficulty of enforcing geographic restrictions in a globalized digital economy where AI tools are integrated deeply into developer workflows.
Frequently Asked Questions
Why is Alibaba banning Claude Code?
Alibaba has classified the tool as high-risk, citing security concerns and potential data vulnerabilities, according to Reuters.
What is Qoder?
Qoder is an internal programming tool developed by Alibaba that the company is mandating as the replacement for Anthropic’s software.
Does Anthropic allow Chinese users?
No. Anthropic prohibits Chinese companies and their foreign-owned entities from using its models and is actively working to close loopholes that allow unauthorized access.
What was the “experiment” mentioned by Anthropic?
According to Thariq Shihipar, Anthropic tested a mechanism in March intended to identify users to prevent account abuse and protect against model distillation.
Are you tracking how your organization manages AI tool security? Share your thoughts in the comments below or subscribe to our newsletter for the latest updates on enterprise AI compliance.
