I think we all know that the messages we type into AI chatbots aren't entirely private. They can be used to train future AI models, for example, and most AI companies have some version of human review in their training and quality-improvement processes. But when I loaded up Gemini today, it hit me with a warning message that felt more urgent:
"Humans review some saved chats to improve Google AI. To stop this for future chats, turn off Gemini Apps activity. If this setting is on, don't enter info that you wouldn't want reviewed or used."
By issuing the warning message, Google is being unusually explicit here. It's making it clear that this is different from simply storing your chat history. It's specifically about using conversations for product improvement and quality control.
I'm not entirely sure why Gemini decided to show me this warning today — I'd not changed anything about my Pro subscription recently — but other users have noticed the same message appearing over the last couple of months.
Anyway, I decided to turn it off, since nobody wants other people reviewing their chats, and that's when I discovered the hidden cost that Gemini asks users to accept.
The privacy tradeoff I wasn't expecting
I clicked through to Gemini Apps Activity to turn the feature off. First, a bit of explanation about why the setting is called "Gemini Apps Activity", because it has very little to do with what most of us would think of as apps.
"Gemini Apps Activity" is really a legacy Google naming convention. It essentially means your interactions with Gemini and connected Google services. Because Gemini can reach into Gmail, Calendar, Drive, and other Google products, Google treats it as part of a broader activity ecosystem rather than simply a chatbot chat log.
Confusing naming convention aside, let's look at what turning the setting off actually does. Here's the kicker: if you disable Gemini Apps Activity, Google will only let you continue conversations if you respond within a 72-hour window. After that, they disappear.
So, imagine you start a conversation with Gemini on Friday evening. Maybe you're planning a vacation, researching a health concern, or working through a complicated career decision. If you don't return to that chat within 72 hours, Gemini won't keep it around. This means that the privacy-friendly option comes with a surprising consequence: your AI assistant stops being able to remember long-running conversations unless you are frequently updating them.
For quick questions, that limitation may not matter. But for the kind of ongoing conversations AI companies like to encourage — planning projects, exploring ideas, or seeking advice over time — it feels like a significant compromise.
ChatGPT does it differently
I don't want to accept that limitation to my Gemini chats, so right now I'm stuck with allowing Google staff to review my chats and use them as training data.
ChatGPT and Gemini both offer ways to limit how your conversations are used, but they frame the choice differently. In ChatGPT, I can switch off Improve the model for everyone and continue using the service much as before. In Gemini, the decision is wrapped up in the broader Gemini Apps Activity setting, where turning it off also means giving up long-term chat history.
At a time when AI companies are pushing users to treat chatbots as personal assistants, therapists, life coaches, and memory systems, Gemini is simultaneously reminding users that some conversations may be reviewed by humans while making it harder to maintain long-term conversations if you opt out.
In ChatGPT, if you select Settings, go to Data Controls, and turn off Improve the model for everyone, you're good to go without any impact on your saved conversations.
With ChatGPT, opting out feels like changing a privacy setting. With Gemini, it feels like making a tradeoff that users shouldn't have to make.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.
