Alibaba has banned employees from using Anthropic’s Claude Code after what’s being described as a hidden backdoor that detects when the tool is being used in China. Staff are being told to switch to Alibaba’s own Qoder instead.
The alleged backdoor reportedly let Claude Code identify Chinese IPs and adjust behavior accordingly. That didn’t sit well with Alibaba, especially given the widening rift between US and Chinese AI companies.
It’s the latest chapter in the ongoing decoupling of AI tooling between the two countries. Chinese firms are increasingly pulling Western AI products from internal use and pushing their own alternatives. Anthropic hasn’t commented on the specific allegations.
