Tracking Conversations: Measuring Content and Identity Exposure on AI Chatbots

We investigate web tracking practices across 20 AI chatbots by analyzing network traffic under controlled conditions, examining whether sensitive conversation data and user identity information are shared with third parties. We find that 17 of the 20 chatbots disclosed information to external entities, with three platforms sharing plaintext conversation snippets via session replay tools, and fifteen platforms exposing conversation URLs or identifiers to advertising and analytics services.

Recommended citation:
Download Paper