Security researcher Thereallo discovered that Anthropic had hidden tracking code in Claude Code using prompt steganography to monitor users in China, flagging timezones, proxies, and connections to Chinese AI labs. Anthropic quickly removed the tracker after the exposure.
Engineer Thariq Shihipar confirmed the code was a March experiment aimed at preventing account abuse by unauthorized resellers and protecting against distillation attacks. The Washington Post reported that unauthorized resellers had been selling access to free models for as little as $1 a month and pro subscriptions for $12.
Comments