Chinese authorities have issued a warning for users to immediately delete the Claude Code AI tool developed by U.S.-based company Anthropic.

The move signals escalating tensions between the U.S. and China over the security and governance of artificial intelligence. Because Claude Code is designed to interact directly with developer environments, any vulnerability could expose critical infrastructure or proprietary corporate data to foreign intelligence services.

Regulators in China reported the security risk on July 8, 2026 [1]. Officials said that specific versions of the tool contain a back-door vulnerability that could allow the software to transmit sensitive information to a remote server [1].

The warning emphasizes the potential for unauthorized data exfiltration, which the regulators described as a significant security risk [1]. While the specific versions of the software containing the flaw were not detailed in the public warning, the directive urged the immediate removal of the tool to prevent further exposure.

This development follows a pattern of increasing scrutiny by Chinese authorities toward foreign AI models. The government has previously restricted various U.S.-based platforms to ensure domestic data remains within national borders, and to promote the use of homegrown AI alternatives.

Anthropic has not yet provided a public response to the specific claims regarding the backdoor. The tool in question is part of a broader suite of AI coding assistants that automate software development tasks—a sector that has become a focal point for cybersecurity concerns in both nations.

“Chinese authorities have issued a warning for users to immediately delete the Claude Code AI tool”

This action reflects the growing 'AI Cold War' where software security is used as a tool for geopolitical leverage. By flagging a 'backdoor,' China is not only protecting its domestic data but also creating a regulatory environment that favors local AI developers over US competitors. This trend suggests that cross-border AI integration will face increasing friction as security audits become more politicized.