Alibaba Bans Employees From Using Claude Code-钛媒体官方网站

TMTPOST — Chinese tech giant Alibaba has issued a ban to prevent its employees from using Anthropic’s Claude Code in workplace environments, starting on July 10.

The move follows internal assessments that flagged the AI coding assistant as “high-risk software” after reports it contained potential backdoors and mechanisms to detect China-linked users

Alibaba’s employees are required to uninstall products related to Anthropic, including multiple series of models such as Sonnet, Opus, and Fable, as well as Agent products including Claude Code. 

Since the beginning of this year, Alibaba had encouraged employees to adopt AI technology by not only introducing free quotas for internal models but also implementing a substantial reimbursement policy for the use of external models. Employees were free to choose external models such as Claude, GPT, and Gemini, with some programmers consuming quotas of up to several hundred dollars each week, frequently using Claude Code, OpenAI Codex, and Alibaba’s Qoder as their Agent tools. This “reverse ban” has cut off access to Claude.

The ban was related to Anthropic’s accusations. On June 10, Anthropic submitted a letter to the U.S. Senate Banking Committee, accusing Alibaba of using approximately 25,000 fake accounts to engage in over 28 million interactions with Claude between April 22 and June 5. This behavior was characterized as “industrial-grade model distillation attacks,” escalating the matter to a national security level.

In response, Anthropic significantly tightened its risk control strategies. Anthropic had previously made similar accusations against DeepSeek, the Moonshot, and MiniMax. In February of this year, Anthropic accused these three Chinese AI firms of conducting large-scale distillation attacks on Claude.

On June 24, Alibaba filed a lawsuit in the U.S. District Court for the Northern District of California, suing the Department of Defense to have it removed from the “Chinese Military Companies List” (the 1260H list).

This list was published by the Department of Defense on June 8, including Alibaba, Baidu, BYD, and other companies. Although the list does not directly equate to formal sanctions, it raised concerns in the market about Alibaba’s ability to use cutting-edge American technology. Alibaba subsequently submitted detailed evidence to prove its innocence but received no response from the Department of Defense.

From late June to early July, Anthropic implemented a new wave of mass account bans on Claude, with many Chinese users being disabled without warning, affecting both personal subscriptions and team accounts. Accounts that were directly paid for through the Anthropic official website and deemed to have violated the rules were not refunded, and the success rate for appeals was extremely low.

Subsequently, some developers conducted reverse analysis and discovered that Claude Code had built-in a “covert trojan” system starting from version 2.1.91, released in April 2026. This system reads the local time zone and checks whether the proxy or custom API addresses contain keywords related to Chinese cloud providers, AI companies, or API proxy service providers, marking Chinese users by replacing punctuation marks in system prompts through steganography. Members of the Claude Code team later publicly acknowledged this “experimental” measure.