China has escalated its scrutiny of Anthropic's Claude Code, the company's AI coding tool, publicly raising concerns about security vulnerabilities and an alleged "backdoor."
According to Yahoo Finance, China says it has found security vulnerabilities in Claude Code. Voz.us and Digitimes report that Chinese authorities have gone further, warning of an alleged "backdoor" risk in the tool — language that implies a hidden pathway for unauthorized access rather than an ordinary software flaw.
The framing is contested. Malay Mail captured the dispute in its headline "Backdoor or bugbear?", signaling that China and Anthropic are clashing over whether the concerns are substantiated or overstated. The reporting describes this as an escalation rather than a one-off complaint, suggesting the friction has been building.
According to UC Today, the security warnings are unfolding against the backdrop of a deepening dispute between Anthropic and Alibaba, the Chinese technology giant. That detail places the security question inside a broader commercial and geopolitical rivalry, where a US AI firm and a major Chinese company are increasingly at odds.
The sources here are headlines and summaries, so the specific technical claims, Anthropic's detailed response, and independent verification of any vulnerability are not spelled out. What is clear is that a national government is publicly challenging the trustworthiness of a widely used AI developer tool.
Why it matters: when a country as large as China labels a popular Western AI tool a security risk, it can shape which software developers and companies are allowed to use — turning a technical dispute into a fault line in the broader US-China technology contest.