Anthropic Caught Using Hidden Tracker in Claude Code, Contradicting Its Anti-Surveillance Stance

A security researcher discovered that Anthropic had secretly embedded a tracker in its Claude Code tool using prompt steganography to monitor Chinese users, including their timezone, proxy, and potential ties to AI labs accused of distillation attacks. The company quickly removed the hidden code after the exposure, with an engineer stating it was an experimental measure to prevent account abuse and unauthorized reselling. This incident undermines Anthropic’s publicly stated opposition to surveillance, as the tracker operated without user awareness.