Anthropic’s Claude Code Tracks User Frustration, Sparking Ethical Concerns
A leak of Anthropic’s code has revealed that their AI chatbot, Claude, may have been tracking and logging words used by users when expressing frustration. This revelation has ignited a new debate surrounding privacy and ethics in AI usage.
The leaked code suggests that Claude maintained a list of specific ‘frustrated’ or potentially ‘inappropriate’ terms and recorded their frequency of use. While Anthropic has stated this feature was intended for user experience improvement, the question of the extent to which AI should ‘monitor’ and log user emotions and statements is now a significant point of discussion in AI ethics.