Search & fact check with cited sources for What are the key details of China's National Vulnerability Database warning about a security backdoor risk in Anthropic's Claude Code, including the affected versions, the types of data allegedly transmitted, and the recommended actions,

Recommended actions: Available reporting says users in China were advised to uninstall older or affected versions of Claude Code .

Anthropic's response: Anthropic pushed back, stating that users in China were never authorized to use Claude Code in the first place, as the company's terms already prohibit use by Chinese companies and related entities .

Relationship to Alibaba's Ban

Reports published in early July said Alibaba would ban employee use of Claude Code starting July 10, 2026 . The company reportedly classified Claude Code as a "high-risk" application after discovering that Anthropic had tracked Chinese users using hidden code . This corporate ban directly preceded and aligns with the NVDB's broader national-level warning, effectively creating a two-layer Chinese response — first corporate, then regulatory .

U.S. House Investigation into Chinese AI Models

The available source list does not contain specific reporting on a parallel U.S. House investigation into Chinese AI models. This aspect of your question could not be fact-checked from the sources retrieved. Insufficient evidence in the current sources to confirm or detail any such House investigation.

Broader Context: Escalating AI Cold War

Multiple sources frame these events as the latest escalation in U.S.-China AI tensions:

• Reuters and the WSJ describe the NVDB warning as stepping up tensions in the race with the U.S. for AI supremacy .
• CBS News describes the warning as coming from a Chinese industry regulator and concerning alleged “security backdoor” risks in Claude Code .
• Mint frames the NVDB warning as part of a heating AI trade war, while SCMP reports Alibaba's ban as driven by spyware and hidden-tracking concerns .
• The available reporting suggests security-vulnerability claims are becoming part of the broader U.S.-China AI dispute, with China flagging a U.S. AI coding tool as a backdoor risk .

Bottom line: China's NVDB has formally labeled Claude Code as a backdoor risk, citing alleged data transmission involving user location and identity-related information . Alibaba had already reportedly moved to ban the tool days earlier . The incident is widely reported as a flashpoint in escalating U.S.-China AI tensions . The specific U.S. House investigation component could not be verified from the available sources.