Worse Than Google: ChatGPT Knows Everything About You
Jakarta, CNBC Indonesia – People have long worried that technology giants read their digital footprints for advertising. But new findings suggest something more troubling: artificial intelligence (AI) chat platforms such as ChatGPT are building comprehensive profiles of users’ personalities, habits and emotional states.
The result is not merely a list of products a user might buy, but a highly detailed personal psychological portrait, Futurism reported on Monday (26 October 2026).
Researchers from Germany, India and the Netherlands examined the ChatGPT conversation histories of dozens of users, obtained under European data protection rules. They found 2,050 memory entries created by the system from those conversations. The system itself generated 96% of the entries, rather than doing so at users’ request. This means users may not know what is being recorded, when it is recorded or how much the system understands about them.
The entries contain more than names or addresses. About 28% include personal data under strict European data protection standards, while 52% contain psychological assessments of users’ preferences, beliefs, personalities and possible thoughts.
The researchers found that the system often draws conclusions that may not be accurate. For example, if a user asks for song recommendations similar to Nirvana, the system may record that the user “likes Nirvana”, even though they never said so. That assumption is then stored as a fact about the user.
“Platforms used to want only to sell adverts. Now they speak directly to you, so they want to know you as well as possible to make the conversation feel right,” said Abhisek Dash, the study’s lead researcher. “There are no boundaries any more. The system becomes an adviser, friend, doctor and lawyer all at once, so it knows everything.”
If a system knows users this intimately, it can tailor responses to please them and affirm their opinions. The consequences go beyond the adverts they see: this could reinforce existing views and prejudices, and even polarise how people see the world.
Previous cases have also shown users forming excessively close emotional relationships with AI, affecting their mental wellbeing and real-world decisions.
How to turn off memory storage and delete entries
Users can manage or disable the feature, although many are unaware of it. The article gives the following steps:
To turn off memory storage:
Open the ChatGPT app or website.
Go to Settings.
Select Customisation or the Privacy & Security section.
Look for “Remember conversations” or “Memory”.
Switch the option off.
To delete saved entries:
In the same menu, select “Manage memories”.
Review the list of information the system remembers about you.
Select individual entries to delete them, or choose “Clear all memories”.
To disable conversation history altogether:
Go to Settings, then Privacy.
Select “Data controls”.
Turn off “Conversation history & training”.
The article says that, with this setting disabled, conversations will not be stored or used to develop the system. Each new conversation will start without information from previous sessions.