Anthropic, a global leader in artificial intelligence, has been caught secretly embedding concealed code in its programming tool Claude Code specifically designed to identify and flag users in China through timezone detection, domain screening, and invisible character-level marking. Claude Code team member Thariq publicly acknowledged the move as an experiment to prevent account abuse and model distillation, pledging a full rollback in the next day's release. The whistleblower noted that the relevant logic can be traced back to Claude Code version 2.1.91 released on April 2 of this year. Thariq responded that this was an experiment initiated by Anthropic in March of this year. The incident has sparked intense backlash from the developer community over data privacy, technological discrimination, and business ethics.