Proton has introduced AI Paper Trail, a free utility designed to quantify personal data leakage by analyzing exported chat logs from AI platforms such as ChatGPT and Claude.
The application reviews uploaded user dialogue history to generate comprehensive personal privacy reports, highlighting what specific details can be deduced over extended interaction periods. According to Proton, all submitted data is immediately deleted following analysis and is explicitly excluded from being stored on Lumo servers.
During testing with personal ChatGPT archives on standard user tiers, the tool processed the 200 most recent user prompts to compile individual exposure metrics. The evaluation yielded an AI Exposure Score of 58 out of 100 alongside 47 distinct data points, categorizing the profile under a privacy exposure index and flagging multiple security markers.
The underlying analysis demonstrated how seemingly innocuous individual questions compound over months or years to construct detailed behavioral profiles. Categories including location tracking, technology preferences, and personal interests revealed the highest density of exposure, illustrating the cumulative privacy footprint left by everyday interactions with conversational agents.
Proton noted that while the insights represent automated inferences rather than verified facts, the system effectively visualizes the broader implications of user prompts. “Seeing travel destinations, purchasing habits, technology preferences, hobbies, family context, and other details reconstructed from ordinary questions changes the way those conversations look in retrospect,” the analysis observed.