Artificial Intelligence

Alibaba Stops Use of Claude Over Chinese User ID Code

Is Anthropic protecting AI innovation from theft or secretly surveilling Chinese users through a hidden backdoor in Claude?

JUL 6·Updated JUL 8

Alibaba Stops Use of Claude Over Chinese User ID Code

Story

Above: An office building at Alibaba Center Shenzhen in China on June 27. Image credit: Cheng Xin/Getty Images

The Facts

Chinese tech firm Alibaba has reportedly banned employees from using Anthropic's Claude Code for work purposes, classifying it as high-risk software with security vulnerabilities. Staff were directed to use Alibaba's own AI coding platform, Qoder, as a replacement.
Tom's Hardware
Reuters
The ban followed the discovery by security researchers that Claude Code contained hidden code to detect China-based users, checking system time zones, proxy servers and network characteristics. The findings were first published in a Reddit post by a user who claimed to have reverse-engineered the tool.
South China Morning Post
TechRadar
Anthropic engineer Thariq Shihipar described the detection mechanism as "an experiment we launched in March" to prevent abuse by unauthorized resellers and protect against model distillation. He said the code was removed on July 1, the day after the Reddit post appeared.
Thariq Shihipar on X
Tom's Hardware
Anthropic accused operators linked to Alibaba's Qwen AI lab of using nearly 25,000 fraudulent accounts to generate 28.8 million exchanges with Claude between April 22 and June 5. The company described the campaign as the largest known model distillation attack against Claude in a letter sent to U.S. Senate Banking Committee leaders.
Reuters
NewsBytes
Alibaba's internal directive reportedly extended beyond Claude Code, with staff instructed to uninstall all Anthropic products, including the Claude Sonnet, Opus and Fable model families. Distillation is the use of more advanced AI models to train less capable ones.
Tom's Hardware
China Daily
Anthropic already prohibits Chinese companies and foreign entities owned by those companies from using its models, maintaining what it describes as the strictest access restrictions of any frontier AI firm toward China. As U.S. AI developers tighten access controls, Chinese firms have increasingly shifted toward domestic and open-source models such as DeepSeek, Qwen and Zhipu.
Reuters
TradingView

Sources Split

LEFT

CENTER

RIGHT

ESTAB
CRIT

CENTER

PRO
ESTAB

The Spin

Pro-China narrative

Anthropic accused Chinese AI firms of distillation, a lawful and normal practice, then embedded covert surveillance code in a developer tool with full filesystem access — disclosed to nobody, and removed only when exposed. This corporate malpractice smear against Chinese firms is laughable now that Anthropic has been proven guilty of what they accuse others of.

Global Times

TNW

Anti-China narrative

Anthropic's detection code existed because Chinese operators were systematically looting its models through stolen credentials and proxy networks — an industrial-scale IP extraction campaign that predated the tracking code by months. The surveillance was a necessary response to secure their product.

Tom's Hardware

Thariq Shihipar on X