
Google Gemini can build a detailed memory of its users over many conversations, boosted by its access to your Google accounts and ability to absorb details from other AI chatbots. Just how much information that includes is suddenly very clear thanks to its new Memory summary feature.
Gemini users can read an overview of what the AI has learned about them, organized into neat paragraphs of potentially disturbing depth. You just have to open Gemini, go to in , and click on the summary option. You can assume it will be information you've directly shared, and anything it has indirectly picked up from your chats. It seems reasonable that users should be able to examine that information, and you can even edit it if there are errors. I tried it out myself, but even knowing what to expect, I was a little taken aback.
The feature presents personal information and professional details, then lays out what it thinks your hobbies and interests are. It gets pretty personal and has everything organized pretty well. That it's just the summary is a reminder that there's probably a lot more that isn't in the short roundup. It's a surprisingly detailed overview of someone who has never deliberately written a biography for the chatbot.
The most interesting thing to look for is whether Gemini has interpreted casual remarks as lasting facts. Mentioning that you're considering learning a language is different from saying you've already mastered it, while asking about a particular hobby doesn't necessarily mean you practice it. The AI makes those distinctions, even if you don't remember doing so in a conversation with it.
I checked to see if there were any errors, and I found some confusion about my actual interests and the experiments I conducted using Gemini. That's the sort of mistake that might seem harmless until it starts affecting unrelated conversations. An assistant that remembers you were researching a particular subject could easily become irritating if it assumes you want every future answer connected to it.
The feature's editing controls are designed to address any such confusion. You can simply tell Gemini the issue, and it will correct the information directly in the Memory summary. Gemini will even ask follow-up questions when a request is unclear. For more direct editing, there's the Memory history option that shows when information was added or modified. That gives users some visibility into how their remembered details have changed. Overall, the AI had a remarkably clear picture of who I am. I didn't necessarily like that idea, but I couldn't deny the fact.
Controlling the profile

Google's privacy documentation makes clear that Memory summary is only one part of Gemini's broader data controls. Removing information from Memory summary doesn't automatically delete the original conversation where you mentioned it. Likewise, turning Memory off prevents Gemini from continuing to learn about you through that feature, but it doesn't automatically erase your conversation history. Anyone concerned about sensitive information should review their Gemini activity settings separately.
I find it tempting to do so myself when I see how much I've exposed to the AI. That said, Google already has a lot of information on me and so many others just from using other apps, so it's hardly a clean slate.
Still, you can always use Temporary Chat to discuss something without adding it to the assistant's ongoing memory. The Memory summary is appealing for those who might not realize how much Gemini remembers about your life. Giving users a clear way to inspect and correct them makes personalization feel considerably less mysterious.
I like the idea of Gemini remembering enough about me to avoid repeating the same explanations every time I open a new conversation. I also want to know when it has misunderstood something and have a straightforward way to fix it. Memory summary offers a welcome degree of control, but ultimately using the AI chatbots means sacrificing some of your privacy to make them useful, just like pretty much any modern online service.


Eric Hal Schwartz is a freelance writer for TechRadar with more than 15 years of experience covering the intersection of the world and technology. For the last five years, he served as head writer for Voicebot.ai and was on the leading edge of reporting on generative AI and large language models. He's since become an expert on the products of generative AI models, such as OpenAI’s ChatGPT, Anthropic’s Claude, Google Gemini, and every other synthetic media tool. His experience runs the gamut of media, including print, digital, broadcast, and live events. Now, he's continuing to tell the stories people want and need to hear about the rapidly evolving AI space and its impact on their lives. Eric is based in New York City.