Gemini Now Lets You Read and Edit What It Remembers About You

Google is rolling out a way to view and change what Gemini remembers about a user, directly from the settings menu. The feature was reported on 18 November 2026. It presents the assistant's stored memory as a document the user can read and modify.
According to Engadget, the memory summary is available to users in the US who hold a paid Google AI plan and have memory turned on. In the Gemini web app, it sits under Settings, then Personal Intelligence, then Memory summary. The summary is a bulleted list sorted into categories such as personal data, career information and interests. An Edit button in the top-right corner lets users rewrite or delete individual lines.
Memory depends on a separate control. Keep Activity determines whether chats with Gemini are saved and linked to a Google Account. Gemini Memory needs it switched on before it can create memories. For users aged 18 or older signed into a personal account, Keep Activity is on by default. If a user turns it off, chats are not saved to the account, disappear after 72 hours, and do not follow the user across devices.
Gemini Memory builds a profile from past chat content, and the assistant uses that profile to personalize its replies. Google's Personal Intelligence feature widens the context further, letting Gemini pull in information from Gmail, YouTube, Search and other Google apps. The memory summary is therefore the visible output of a profile whose inputs reach beyond the chat transcripts.
The broader context here is that personalization in large language model (LLM) products has largely operated out of sight. A model that adapts to a user's career, interests or household is only as transparent as the interface that exposes what it has retained. A readable, editable list is a different kind of control from a privacy policy or a blanket data deletion. The user can inspect a specific claim the assistant has made and correct it.
The more practical question is what an edit does to the underlying profile. The Edit button changes lines in the summary, but users will want to know whether that change carries through every place Memory draws on, including the connected-app context that Personal Intelligence adds. People evaluating the feature for work accounts or shared devices will need that answer more than the menu path.
The rollout is also narrow for now. Access is limited to US paid-plan users with memory enabled, so these controls are not yet available to every Gemini user. Everyone outside that group continues to see the existing memory behavior, whatever their Keep Activity setting.
The interaction between the two toggles is the detail most likely to catch users out. Turning Keep Activity off removes the persistence that Memory relies on. It also removes cross-device continuity. A user who disables it to limit retention will lose the personalization the summary exposes. One switch touches saved history, memory creation and device sync at once.
For the people who build on or evaluate these systems, the change is a useful template. Inspectable, user-editable memory is a concrete design pattern. It turns personalization from an opaque inference into a document with line items. That does not settle questions about retention or cross-service data flow. It does give users a handle on the assistant's model of them.
In my view, that kind of handle is what makes adaptive software easier to live with.


