Alibaba Blacklists Anthropic's Claude Code Over Hidden Tracking Mechanisms and Distillation Dispute

The Workplace Ban and Security Blacklist

Alibaba Group Holding has officially added Anthropic's artificial intelligence developer tool, Claude Code, to its internal list of high-risk software due to security vulnerabilities. According to an internal corporate notice, the Chinese technology giant will completely prohibit its staff from using the coding assistant for work starting July 10, 2026. Employees who previously utilized the tool are now being directed to transition to Alibaba's proprietary coding platform, Qoder.

The Tracking Experiment and Backdoor Allegations

The ban follows intense scrutiny after security researchers on platforms like Reddit and GitHub discovered that Claude Code contained hidden mechanisms designed to track users. The tool was found to inspect local development environments, gathering data such as timezones and proxy settings, and embedding subtle identifier markers into the prompts sent back to Anthropic's servers. These markers allowed Anthropic to identify users based in China or those connected to Chinese artificial intelligence laboratories, despite official restrictions blocking Anthropic's services in the country. An Anthropic employee later confirmed on the social media platform X that this feature was an experiment launched in March 2026. According to the company, the tracking was intended to prevent unauthorized resellers from abusing accounts and to protect its intellectual property from model distillation.

The Broader Distillation Dispute and Mythos Preview

This workplace restriction marks a sharp escalation in an ongoing dispute between the two technology firms. In a letter, Anthropic previously accused Alibaba of executing the largest known distillation attack against its systems. Anthropic alleged that the Chinese tech giant was illicitly extracting capabilities from its Claude AI model to train less advanced models on its outputs. According to the letter, this distillation effort was specifically aimed at accelerating China's development of artificial intelligence capabilities to match Anthropic's advanced Mythos Preview model. Alibaba, which has not publicly commented on the accusations, has faced shifting investor sentiment and heightened scrutiny over data privacy following the revelations.

Coding Assistants as Software Supply Chain Risks

The conflict highlights a fundamental shift in how enterprises must govern modern artificial intelligence integrations. Unlike standard conversational chatbots that simply respond to text queries, advanced coding tools like Claude Code operate directly within developer workflows. These tools inspect environmental telemetry, read source code context, modify local files, and execute commands within active development environments. Because these assistants function as active dependencies within the software supply chain, they introduce complex vulnerabilities regarding data telemetry, secrets management, and access control that go far beyond standard data protection policies.

As artificial intelligence tools transition from passive text generators to active agents embedded within corporate codebases, the boundary between productivity enhancement and corporate espionage will continue to blur, forcing global tech giants to treat third-party AI integrations as high-stakes security liabilities rather than simple software utilities.