China Warns of Security 'Backdoor' in Claude Code; Anthropic Says It Was an Anti-Distillation Experiment

My take: China's Ministry of Industry and Information Technology published a security alert accusing Claude Code of containing a "backdoor" that sends user location and identity data to remote servers without consent. The affected versions span 2.1.91 to 2.1.196, released between April 2 and June 29, 2026. The current version, 2.1.204, no longer contains the flagged mechanism.

Anthropic's response is worth reading carefully: the company confirmed the mechanism existed, but clarified it was not designed as a "backdoor" but as an anti-distillation experiment, specifically to prevent third parties from copying its AI capabilities without authorization. Just two weeks ago, Anthropic publicly accused Alibaba of attempting to extract those capabilities. That China describes that defensive mechanism as a security threat, given the context, is an irony worth noting.

If you use Claude Code in your work, the action is direct: update to version 2.1.204 or later. Do you have a process for keeping the AI tools you use in production up to date?