AI chatbots are learning more about you than you might expect. Discover what a new study reveals about chatbot memory, hidden personal profiles, and the privacy risks you shouldn’t ignore.

AI chatbots are becoming more personal, but a new study raises an important question: how much do they really know about their users? A study conducted on the memory system of ChatGPT revealed that 96% of memories stored within the system had been formed without the user making a request for the bot to remember something. This has raised fears regarding privacy and data collection by artificial intelligence from people’s conversations.

AI Chatbots Are Building Detailed User Profiles

The research, titled “The Algorithmic Self-Portrait: Deconstructing Memory in ChatGPT,” was based on 2,050 memory entries obtained from 80 people who allowed the researchers to access their data in ChatGPT. The research team had scientists from the Max Planck Institute for Software Systems and the Max Planck Institute for Security and Privacy, Germany.

The researchers found that 52% of the memories contained insights into users’ psychological states, including their emotions, beliefs, intentions, and desires. Another 28% included personal information covered by the European Union’s General Data Protection Regulation (GDPR).

These findings suggest that chatbots may retain more than basic preferences. They can also build a picture of how users think, feel, and make decisions.

Why Does Chatbot Personalization Create Privacy Risks?

Personalized responses will make the chatbot more useful. The bots are able to recall past conversations, understand the preferences of the user, and offer personalized responses. But this could pose a danger since users are likely to divulge sensitive information.

Lorrie Cranor, director of Carnegie Mellon University’s CyLab Security and Privacy Institute, explained the difference between talking to a person and interacting with AI.

“Humans will sit next to a stranger on a plane or a bus or a train and tell them things they would never tell people that they’re going to see again,” Cranor said. She noted that, unlike a stranger, an AI system records conversations.

Abhisek Dash, who led the research team, has also stated that chatbots are becoming advisors when it comes to issues of health, finance, employment, and many other personal matters. Information provided in one area might affect advice in another area.

The risks extend beyond data leaks. Depending on the service and its settings, conversations may also be used to improve AI models or personalize advertising. OpenAI says advertisers cannot access users’ ChatGPT conversations or memories, and ads do not influence chatbot responses.

How Can Users Protect Their Personal Information?

A user can limit the amount of information disclosed through the use of general questions rather than disclosing any personal information. For instance, a person looking for information on diabetes can ask what a diabetic patient needs to look out for, instead of sharing personal medical records.

ChatGPT also provides memory controls, including options to review or disable memory and use Temporary Chat. Users can separately manage whether their conversations are used to train future models.

According to Business Fortune, the AI firms might require being more transparent regarding how they will save the sensitive information and get permission from the users before saving such information. With the increasing involvement of chatbots in decision-making processes in our lives, the future of AI technology will not only be about intelligent responses but also about privacy.

FAQs

What did the ChatGPT memory study discover?

Researchers found that 96% of the stored memories examined were created without users explicitly requesting them, raising concerns about transparency and control.

What is an algorithmic self-portrait?

It is a profile built from information users share with AI chatbots, including personal details, preferences, emotions, and inferred psychological characteristics.

Can ChatGPT remember sensitive personal information?

Yes. Depending on its memory features and settings, ChatGPT may retain information from past conversations, including personal preferences and other details users have shared.

How can users limit the personal information AI chatbots collect?

They can avoid unnecessary identifying details, review memory settings, use Temporary Chat where available, and manage whether conversations are used for model training.

Will AI chatbots become more privacy-friendly in the future?

They could, particularly if companies introduce clearer consent processes, better memory controls, and greater transparency about how personal information is stored and used.

 

Sources: https://gizmodo.com/gizmodo-is-launching-a-daily-newsletter-about-ai-2000829171 http://tech.yahoo.com/ai/chatgpt/articles/inside-chatgpt-ai-builds-portrait-161134976.html https://gizmodo.com/ai-chatbots-are-building-an-algorithmic-self-portrait-of-you-2000775529