Gemini Memory summary shows how much Google's AI remembers
Google's Gemini now offers a Memory summary that lists what the assistant has learned about a user. TechRadar's reviewer found it detailed, personal and sometimes wrong, with editing tools to correct it.

Key takeaways
- Gemini's Memory summary lists personal, professional and hobby details the assistant has collected.
- Users can reach it through Settings, then Personal Intelligence, and edit wrong entries directly.
- Memory history shows when details were added or changed.
- Deleting a summary item or turning Memory off does not erase the original chats.
In this article
What the Memory summary shows
Google's Gemini assistant can build a picture of a user across many conversations, helped by access to their Google accounts and by details it picks up from other AI chatbots. According to TechRadar, a new Memory summary feature now lets users read an overview of what the assistant has learned about them.
The summary is arranged in paragraphs and covers personal information, professional details and what Gemini believes a person's hobbies and interests are. TechRadar's reviewer described the overview as surprisingly detailed for someone who had never written a biography for the chatbot, and noted that the short roundup probably represents only part of what is stored.
How to find and edit it
The feature sits in Gemini's settings. Users open Gemini, go to Personal Intelligence and select the Memory summary option, according to TechRadar. The page shows what the assistant has taken from conversations and from other sources linked to the account.
Editing controls are built in. A user can tell Gemini that a detail is wrong and the assistant will correct the summary directly, and it may ask follow-up questions when a request is unclear. A separate Memory history option shows when information was added or changed, giving some visibility into how the remembered details have evolved.
Where the memory can go wrong
TechRadar's reviewer found errors in the summary, including confusion about real interests and about experiments run with Gemini. The report notes that such mistakes can seem harmless until they affect unrelated conversations, because an assistant that remembers a past research topic may keep linking later answers to it.
The reviewer also highlighted a subtler risk: Gemini may treat a casual remark as a lasting fact. Saying that someone is considering learning a language is not the same as saying they have mastered it, and asking about a hobby does not prove that a person practises it. The report says the assistant does try to make those distinctions.
Privacy settings users should check

Google's privacy documentation makes clear that the Memory summary is only one part of Gemini's data controls, according to TechRadar. Deleting an item from the summary does not automatically remove the original conversation where it was mentioned. Turning Memory off stops Gemini from learning more through that feature, but it does not erase existing conversation history.
People worried about sensitive details should review their Gemini activity settings separately, the report says. Gemini also offers Temporary Chat, which allows a discussion without adding it to the assistant's ongoing memory. TechRadar frames the summary as a useful way to inspect and correct what the assistant knows, while noting that using such chatbots still means giving up some privacy.
Sources
Spotted a mistake? See our corrections policy. Read how News by OnDevice works in our editorial policy.



