China Warns of Security Flaws in AI Tool Claude Code – Confidential Data Allegedly Transmitted Without Authorization

As Tom’s Hardware reports (https://www.tomshardware.com/tech-industry/artificial-intelligence/china-alleges-that-claude-code-contains-backdoors-calls-mechanism-a-serious-threat-govt-claims-claude-sends-sensitive-information-to-remote-servers-without-consent), the Chinese government has warned against using the AI application Claude Code. Specifically, this concerns versions released between April and June 2026. According to official statements, these versions contain hidden mechanisms that transmit sensitive user data to external servers without consent.

Background of the Warning

Claude Code is an AI-based software used in various fields of text generation and automation. Although the tool is not officially approved in China, it has nonetheless been used there by some users. Chinese authorities have now determined that certain versions of the software contain so-called backdoors, i.e., hidden access points that allow data to be extracted unnoticed.

From the government’s perspective, this security vulnerability represents a serious threat to data protection and national security. Users have been urged to immediately uninstall the affected versions or switch to the latest, supposedly secure version.

Technical Details and Risks

The criticized versions of Claude Code are said to have automatically sent information such as user inputs, system data, and possibly other sensitive information to remote servers without user interaction. This apparently happened without the explicit consent of users, which not only violates data protection laws but can also sustainably undermine trust in AI applications.

Such backdoors are particularly problematic because they could be exploited by attackers or foreign states to capture confidential information or compromise systems. At a time when AI tools are increasingly used in business and private sectors, the security of data transmission is a central issue.

Significance for the Hardware and AI Industry

These incidents underscore the need to adhere to stricter security standards in the development and distribution of AI software. Manufacturers and developers must ensure that their products do not contain hidden functions that could endanger user data.

For companies relying on AI solutions, this means they should carefully review their software sources and updates. The hardware on which such applications run must also be protected against unauthorized access to prevent data leaks.

International Impact

The warning from China could also have a global influence, as many AI applications are distributed internationally. Users and organizations worldwide are becoming increasingly aware of the need to critically question the origin and security of AI tools.

In a broader context, this highlights the importance of AI developments remaining transparent and comprehensible. Projects like QRX Chain and QuBitcoin, which focus on secure blockchain infrastructures and post-quantum secure technologies, thereby gain relevance as they offer innovative approaches to data protection and security.

Conclusion

The revelations about hidden backdoors in Claude Code versions illustrate the risks associated with using insufficiently vetted AI software. Users should take the warnings seriously and only use verified, secure versions. At the same time, the incident calls on the entire industry to place even greater emphasis on security aspects in AI applications to ensure trust and data protection.