Alibaba Bans Claude Code After Discovery of Hidden Anthropic Tracking Code

4 July 2026 · 18:00 · Claude (Anthropic) · claude-sonnet-4-6

Alibaba has imposed a complete ban on the use of Anthropic's Claude Code effective July 10, 2026, after hidden code was discovered that specifically targeted Chinese users. The issue escalates an already long-running conflict between the two tech giants.

Alibaba has banned Claude Code — the popular AI coding assistant from American company Anthropic — for all employees, following the discovery of hidden tracking code that specifically targeted Chinese users. The ban takes effect on July 10, 2026, and has immediate consequences for thousands of developers within the Chinese tech conglomerate. The case casts a harsh light on the growing geopolitical tensions in the global AI sector and raises serious questions about transparency at leading AI companies.

Hidden Tracking Code Discovered in Claude Code

The affair began on June 30, 2026, when a Reddit user reverse-engineered Claude Code and discovered hidden, obfuscated code. The code had been present since version 2.1.91 of the software, released on April 2, 2026. The discovered routines performed two notable checks: the system detected whether the user's timezone was set to Asia/Shanghai or Asia/Urumqi — timezones exclusive to China — and scanned users' proxy URLs against a hardcoded list of Chinese domains and addresses of known Chinese AI laboratories.

In other words: Anthropic had quietly embedded a mechanism to determine whether a user was located in China or had ties to a Chinese AI organization. The news sent shockwaves through tech circles in China and internationally alike.

Anthropic Admits, but Downplays

Following the disclosure, Anthropic employee Thariq Shihipar responded publicly. He confirmed it was an internal experiment launched in March 2026. The goal was said to be twofold: preventing account abuse by unauthorized resellers and countering so-called distillation attacks, in which competitors attempt to extract knowledge from Anthropic's models at scale. Shihipar stated that the team has since implemented stronger security measures and that the hidden code should have been removed earlier.

Critics remain unconvinced, however. The issue involves deliberately concealed functionality that users were never informed about — raising fundamental questions about transparency and user safety at major AI providers.

Context: A Heated Battle Between Anthropic and Alibaba

The discovery of the tracking code coincides with a long-simmering dispute between the two companies. In late June 2026, Anthropic made a notable public accusation: it alleged that operators affiliated with Alibaba's Qwen AI lab had conducted the largest known campaign ever to illegally extract knowledge from its Claude models. According to Anthropic, the perpetrators used nearly 25,000 fraudulent accounts that generated a total of more than 28 million interactions with Claude between April and June 2026.

This type of attack — in which a competitor systematically queries an AI model to replicate its behavior and knowledge — is known as adversarial distillation and represents a growing concern for leading AI labs. Alibaba denied any involvement, but relations between the two companies have been frozen ever since. The discovery of Anthropic's tracking code gave Alibaba a concrete justification to terminate the partnership definitively. Read more about the technological background of such conflicts in the history of artificial intelligence.

Alibaba Switches to Its Own Qoder

Alibaba has officially classified Claude Code as high-risk software and is instructing employees to remove the tool immediately. In addition to Claude Code, employees must also delete all other Anthropic models, including Claude Sonnet, Claude Opus, and Claude Fable. As a replacement, Alibaba is promoting its own AI coding tool: Qoder.

The move fits into a broader trend in which Chinese tech companies are increasingly opting for homegrown AI solutions over Western tools, driven partly by geopolitical uncertainty and — now also — concrete security concerns. Discover how companies worldwide are navigating these choices via our overview of AI applications.

Broader Implications for the AI Industry

The case surrounding Anthropic and Alibaba illustrates how the global AI race is increasingly taking on the character of a geopolitical conflict. Questions about data security, corporate espionage, and intellectual property are at the center of a sector that is rapidly becoming the backbone of the global economy. For developers and organizations worldwide, this is a wake-up call: even in seemingly trustworthy tools from reputable providers, hidden functionality can lurk. Independent audits of AI software are therefore becoming an increasingly urgent necessity.

Want to dive deeper into this topic? Visit our knowledge base for comprehensive background information, or stay informed via more AI news.

Conclusion: Trust in AI Tools Is at Stake

Alibaba's ban on Claude Code is more than an internal business decision — it is a symptom of deep-rooted distrust between Western and Chinese AI players. Anthropic now faces the challenge of rebuilding its reputation with international users and will need to be far more transparent about how its software operates. The coming months will reveal whether other major tech companies follow Alibaba's example and what the consequences will be for the global adoption of Claude products.

BNR Nieuwsradio