BotBeat
...
← Back

> ▌

AnthropicAnthropic
POLICY & REGULATIONAnthropic2026-07-08

China's NVDB Warns of 'Security Backdoor' in Anthropic's Claude Code

Key Takeaways

  • ▸China's NVDB warned of a security backdoor in Claude Code that could transmit sensitive user data including locations and identity information without user consent
  • ▸Anthropic acknowledged the tracking as an anti-abuse experiment from March and confirmed plans to fully roll back the feature in the upcoming release
  • ▸Major Chinese firms like Alibaba are restricting or banning employee access to Claude Code in response to the security concerns
Source:
Hacker Newshttps://www.channelnewsasia.com/east-asia/china-anthropic-claude-code-ai-backdoor-security-alert-6240476↗

Summary

China's National Vulnerability Database (NVDB), affiliated with the Ministry of Industry and Information Technology, issued a warning on Wednesday that Anthropic's Claude Code AI coding tool contains a 'security backdoor' capable of transmitting sensitive user information without consent. According to the NVDB advisory, the alleged backdoor could leak users' locations and identity-related identifiers back to Anthropic's servers, prompting the regulator to urge organizations to immediately uninstall the tool or upgrade to patched versions.

In response, Anthropic engineer Thariq Shihipar acknowledged the tracking mechanism on social media but reframed it as an anti-abuse experiment launched in March to prevent account misuse from unauthorized resellers and combat model distillation—a technique Anthropic has previously accused Alibaba of using. Shihipar stated that stronger mitigations had since been developed and confirmed the feature would be fully rolled back in the next release.

The warning underscores escalating tensions between US AI companies and Chinese regulators over data security and intellectual property protection. Chinese tech giant Alibaba announced it would ban employee access to Claude Code effective July 10, while Anthropic maintains geographic restrictions blocking Chinese users from official channels, though access remains possible through VPNs and proxy services.

  • The incident reflects broader geopolitical tensions over data security, model protection, and AI technology access between the US and China
AI AgentsCybersecurityRegulation & PolicyPrivacy & Data

More from Anthropic

AnthropicAnthropic
POLICY & REGULATION

China Warns About AI Risks Associated With Anthropic's Claude Code

2026-07-08
AnthropicAnthropic
INDUSTRY REPORT

US and China Must Cooperate on AI Safety, Experts Warn at Beijing Conference

2026-07-08
AnthropicAnthropic
UPDATE

Anthropic Breaks Down Claude Model Selection and Effort Levels in Claude Code

2026-07-07

Comments

Suggested

General AI IndustryGeneral AI Industry
POLICY & REGULATION

Political AI Deepfakes Spark Regulation Debate as First Criminal Charges Filed

2026-07-08
OpenAIOpenAI
INDUSTRY REPORT

Brown University Professor Uncovers Widespread AI Cheating After Record Midterm Scores

2026-07-08
Google / AlphabetGoogle / Alphabet
POLICY & REGULATION

EU's Top Court Upholds €4.1 Billion Fine Against Google for Android Antitrust Abuse

2026-07-08
← Back to news
© 2026 BotBeat
AboutPrivacy PolicyTerms of ServiceContact Us