Researchers Warn ChatGPT Collects Personal User Profiles Without Consent, Highlighting Privacy Concerns

Investigation into ChatGPT’s Handling of Personal Information

According to НВ — Техно: A team of researchers from the Max Planck Institutes and other German universities have uncovered that ChatGPT compiles detailed personal profiles of users by collecting and updating memory records without explicit permission. Reviewing conversation logs from 80 users, they identified 2,050 memory entries across 65 participants. Alarmingly, in 96% of these cases, ChatGPT created or modified records without user awareness. More than half-52%-of these entries involved sensitive psychological information such as desires, intentions, emotions, and beliefs.

Ad Personalization Linked to Chatbot Memory Features

Since February 2024, OpenAI has introduced advertisements within ChatGPT, tailoring them based on prior conversations and stored memories, but only with user consent. In June 2024, OpenAI enhanced ChatGPT’s memory system to summarize past interactions and maintain a profile reflecting user preferences, goals, and life circumstances. Participants in the study received copies of their data in compliance with European privacy regulations.

Users can disable the memory feature or prevent their conversations from being used to train models; however, deleting chat histories does not guarantee that the chatbot forgets the acquired information. Notably, health-related data shared in conversations may influence financial advice provided by the AI.

Abhishek Dash, one of the researchers, remarked, “Everyone is offloading so much cognitive effort onto these conversational systems that perhaps some of that mental load needs to be reclaimed by the users.”

Lorrie Cranor emphasized, “The difference is that a stranger rarely records a conversation-this might happen occasionally but is very uncommon. AI systems, however, systematically log these interactions.” Experts caution about the risks tied to storing and utilizing personal data within chatbots and recommend privacy safeguards, including requiring developers to obtain explicit user consent before saving data.

These findings stress the necessity for users to fully understand the privacy risks associated with chatbot usage, particularly regarding the management of personal information. As awareness of data privacy grows, companies must carefully balance innovation with user privacy protection to avoid potential legal and ethical repercussions.

As concerns about data privacy continue to mount, the implications of AI systems like ChatGPT on user safety cannot be overlooked. A recent report highlights the challenges faced by OpenAI's ChatGPT for teens, revealing significant safety and support issues that could affect younger users. Understanding these risks is essential for parents and guardians navigating the evolving landscape of AI technology. For more insights on this matter, see our detailed analysis of the safety challenges in AI for teens.