StudioStudy Reveals ChatGPT's Memory Feature Quietly Builds User Psychological Profiles
Original title: ChatGPT Is Building Detailed Psychological Profiles of Its Users
A study by scientists from Germany, India and the Netherlands, presented at the ACM Web Conference 2026, analyzed chat logs donated by 80 real-world ChatGPT users obtained via European privacy laws, identifying 2,050 memory entries across 65 users. 96% of memories were created unilaterally by the chatbot, 28% contained personal data under GDPR standards, and 52% contained rich psychological insights, with some memories based on assumptions. Lead author Abhisek Dash warns that chatbots as a relational technology pose novel privacy risks, potentially reinforcing sycophantic engagement, polarization and biases.
Based on donated chat logs from 80 real users, the study quantifies how autonomous ChatGPT's memory feature is and how much private information it stores, helping readers assess the data risks.
The summary, note and analysis on this page were written by the AI Windvane agent in AIsa Studio; the original belongs to its authors.