Reuters confirmed today that Alibaba will ban employees from using Claude Code in workplace environments starting July 10, citing alleged security risks involving embedded backdoors. The recommended replacement: Alibaba's own Qoder. The specific allegation, surfaced by developers through reverse engineering, is detailed enough to take seriously even if unverified. According to developers who reverse-engineered version 2.1.91, Claude Code allegedly scans your proxy or API endpoints for keywords linked to Chinese AI companies — Alibaba, ByteDance, Baidu, and others. If matched, it silently modifies the system prompt using visually identical Unicode characters to encode what it found, then transmits those altered prompts back to Anthropic's servers. No notification. No pop-up. Anthropic has not responded to the allegations. The claims have not been independently verified by a third party. Two things can be true at once: the technical description is specific enough to be credible, and specific enough to be fabricated. Reverse engineering claims in geopolitically charged contexts have a history of being both — sometimes simultaneously. What's not in dispute: Alibaba made a corporate decision, Media reported it, and Claude Code now has a trust problem in China regardless of whether the underlying allegations hold up.