NewsBite analysis · AI, Security
The Hindu
China issues 'backdoor' security alert over Anthropic's Claude Code
A warning has been issued by China's cybersecurity platform regarding a serious backdoor risk found in specific versions of Anthropic's AI coding tool, Claude Code. The alert advises organizations to review their affected systems and to consider either uninstalling the tool or upgrading to a safe release. Additionally, the warning has prompted Alibaba to ban the use of Claude Code among its employees due to concerns about features that can identify users connected to China. Anthropic clarified that the so-called backdoor is merely an experimental feature designed to prevent abuse and asserted that access to Claude is restricted in China.
MonitorConfidence: HighShort term horizonCybersecurity risk: High
Published 9 Jul 2026, 06:35 UTC1 min read
Visit original article ↗Lead image — The Hindu
Summary
TL;DR
- China's cybersecurity platform warns of a backdoor risk in Claude Code.
- Affected versions are 2.1.91 through 2.1.196.
- Organizations advised to uninstall or upgrade to secure releases.
- Alibaba has banned the use of Claude Code at work.
- Anthropic claims the backdoor is an anti-abuse mechanism.
Interesting facts
- This security alert highlights international scrutiny of AI coding tools.
- Alibaba's response indicates heightened concerns over AI tools linked to national security.
Coverage gaps