Alibaba has prohibited its employees from using Anthropic claudio Code coding wizard starting July 10, 2026, after a developer discovered a hidden tracking code that identifies users in China, according to The Next Web.
The company has added Claude Code to its internal list of high-risk software and advises its employees to switch to Qoder, its own coding platform.
Anthropic engineer Thariq Shihipar confirmed on X that the tracking is part of an experiment launched in March and describes it as a measure to prevent account abuse and model distillation. The monitoring mechanism has been active since April.
Why Alibaba Banned Claude Code
In late June, a developer analyzing Claude Code discovered hidden markers that check the system’s time zone, scan proxy URLs against Chinese domains and AI labs, and send signals within system prompts to Anthropic’s servers.
Because Claude Code requires deep access to the local file system to modify and execute code, these tracking mechanisms raised concerns about cross-border data transparency and compliance within the developer community.
Chinese cybersecurity company Huorong Security noted that such mechanisms pose operational and legal risks for companies operating in China and Hong Kong.
Anthropic engineer Thariq Shihipar explained on X that this tracking was part of an experiment launched in March 2026. He stated that the system was designed to prevent account abuse by unauthorized resellers and protect Anthropic models from distillation.
Distillation is a process in which the results of a powerful AI model are used to train a smaller, less expensive one. Anthropic has actively opposed distillation, arguing that it threatens the business models of frontier AI companies.
In June, Anthropic reported to the US Senate Banking Committee that entities linked to Alibaba’s Qwen AI lab had carried out a large-scale distillation campaign against Claude using thousands of fraudulent accounts. Alibaba has denied these allegations.
Compliance concerns for financial sectors and what this means for users in China
Monitoring practices are especially important for companies and financial institutions operating in Hong Kong and mainland China. In both regions, regulators strictly enforce data sovereignty rules, making the collection and tracking of hidden data a serious compliance issue.
Companies that need to control what data leaves their on-premises systems may face legal risks if they use undisclosed tracking mechanisms, even if the software they use provides useful features. The Alibaba ban illustrates this risk assessment on a larger scale.
For developers and organizations using Claude Code in China or Hong Kong, this discovery raises several practical questions:
- Does current use of Code Claude comply with internal data sovereignty policies and local laws?
- What information is sent to Anthropic servers during normal operation?
- Is the purpose of monitoring, as stated by Anthropic, acceptable within the organization’s risk limits?
- Are there domestic alternatives, such as Alibaba’s Qoder, or tools developed by DeepSeek, Qwen and other Chinese AI vendors?
- Outside China, users are less directly affected by the ban, but should be aware that similar controls could be part of global systems, particularly those used by Anthropic for abuse detection.
A broader shift away from US AI tools and what comes next
The controversy has accelerated a broader effort by Chinese technology companies to reduce dependence on American artificial intelligence software. More and more companies consider that foreign tools carry unpredictable regulatory and access risks.
Lizzi Lee, a fellow at the China Analysis Center at the Asia Society Policy Institute, said the conflict highlights how technological competition has shifted toward access control and sovereignty.
He explained: “If a US AI encryption tool can detect Chinese usage or proxy access, it is understandable that major Chinese tech companies would prefer that their employees not use it internally.”
This restriction provides Alibaba with additional reasons to promote its own Qwen models in different business units. By limiting external tools, the company is strengthening its domestic AI ecosystem and encouraging staff to adopt proprietary platforms.
Alibaba’s decision reflects a broader trend of restricting artificial intelligence tools that cross national borders. Recently, the US government ordered Anthropic to suspend access to Fable 5 and Mythos 5 for foreign citizens; The restrictions were later eased following negotiations with the Department of Commerce.
OpenAI also delayed the public release of GPT-5.6 at the request of US authorities. Now, a large Chinese tech company is limiting Anthropic’s encryption tool for its employees due to sovereignty and tracking concerns.
This pattern shows that AI companies operating globally face increasing pressure from governments and enterprise customers to disclose tracking mechanisms, control data flows, and provide options that meet local compliance standards.
Alibaba’s ban will take effect on July 10, 2026. Anthropic has not yet said whether it will disable, fully disclose, or modify its tracking features in response. Users in China should watch for further statements from Anthropic and guidance from local regulators on permitted data collection practices for AI development tools.
Users can stay informed by monitoring Anthropic’s status page and official communications for updates on the future of the tracking mechanism. Chinese developers can still access Claude Code through personal accounts, although restrictions imposed by Alibaba limit use within the company.
