Product

Gemini Now Lets You Keep Your Chats and Still Say No to AI Training

One switch used to decide two things at once, and now it decides one. The second change runs the other way and keeps more of what you assumed was gone.

Emily CarterEmily Carter
Heat: 1,500
Chat bubbles sealed inside a vault door with a retention dial and two faded 72-hour timers

Dek: One switch used to decide two things at once, and now it decides one. The second change runs the other way and keeps more of what you assumed was gone.

What the two Gemini history settings now control

Google split a single control into two this week, and the split is the part that matters. Until now, one Keep Activity switch governed both whether Gemini saved your conversations and whether Google could train its models on them. Turn it off and the chat history went with it. Leave it on and you accepted both halves of the deal.

Now they run separately. You can keep a conversation in your history and still refuse training. That's a real change, not a relabelling of the same toggle.

What each one does, in plain terms. Chat history is your own record inside the app, the thing you scroll back through on a Wednesday to find the itinerary you built on Monday. The training setting decides whether those conversations become material Google can learn from. Keeping the first no longer implies the second.

One detail that gets lost: history off and temporary chat aren't a delete button. Both are ways of saying "don't keep this in my list", and after this week both end up in the same place on Google's side, inside a two-month window.

Gemini chat retention: why 72 hours became about two months

The second change moves the other way, and it's the one to read twice. Chats made with history off, plus temporary chats, used to be held for 72 hours, according to Google's Gemini Apps Privacy Hub. An archived copy of that page from 12 November still says so. The new figure is about two months, and it's rolling out globally in the Gemini app over the coming weeks.

Where did the three-day figure come from? Google's own privacy documentation, which stated it plainly until this week. Nobody outside the company set that number, and nobody outside it can say why the number changed.

So the conversations people actively chose not to save now sit on Google's servers for roughly twenty times longer than before. Google's stated reasons: responding to users, processing feedback, protecting its services and users, and handling potential system failures. Its privacy hub also says temporary conversations aren't used to train models.

Both things can be true at once. Holding data for support and debugging is ordinary practice, and two months is still the longest window among the three big assistants.

Your default depends on Keep Activity, not on your location

If you had Keep Activity switched off before the change, both new settings start off. Nothing was quietly reset to on, which is the fear most people bring to a privacy announcement.

If you had it on, chat history stays on and the training decision becomes yours, separately. The starting point is the state you already picked, so the fair reading is that the split adds a choice rather than removing one.

The practical test takes ten seconds. If your Gemini history still shows conversations from last month, Keep Activity was on for you, and the training half is a question you haven't answered yet.

Google's description doesn't mention regional differences in these defaults. A reader in California exercising CCPA rights and a reader in London covered by UK GDPR should see the same two switches, with the divergence showing up in how you file deletion or access requests, not in what you get on day one.

Google Gemini privacy compared with ChatGPT and Claude

On retention, Google is now the outlier. OpenAI says it may keep temporary ChatGPT chats for up to 30 days. Anthropic holds incognito Claude conversations for 30 days by default. Two months is longer than either of those windows.

None of the three offers hard deletion on a timer, and all of them carve out legal holds, safety investigations and reviews of policy violations. That's why the honest comparison is between stated windows rather than guarantees.

No delete button anywhere.

The training side is where the change helps most. The old coupling forced a choice between your own records and your own data, which is a bad trade for anyone who uses Gemini as a working notebook. Untangling it removes that trade.

What Google won't say about the Gemini chat retention change

Google didn't answer what prompted the change, or whether it affects how conversations are accessed and reviewed. That's the gap worth naming, because a retention extension could be housekeeping or it could reflect how support and safety teams work. The company hasn't said which.

Regulators may want a word. Extending retention under a safety rationale is exactly the sort of change European data protection authorities have treated as needing fresh notice, and the UK's Information Commissioner's Office has pushed back on retention periods that outlive their stated purpose more than once.

How to check your Gemini AI training opt out today

Open the Gemini app and go into its settings. Look for two controls where there used to be one, and treat them as separate decisions: history for convenience, training for consent.

If you want the Gemini AI training opt out in place, switch training off and leave chat history alone. That combination wasn't available last week.

Then check two follow-ups. Whether your temporary chats still behave the way you expect, given the window behind them changed. And whether your employer's Gemini agreement inherits the two-month figure, which is the question most IT teams will ask before they sign anything.

Worth watching next is whether the two-month number shows up in enterprise terms and outside the consumer app, because that's when a consumer setting turns into a contract clause.

Share This Story

Sources

Related AI News

A search bar with an assistant sidebar sliding in from the left and shortcut chips underneath
Product

Google Put Your Chat History on the Search Homepage

The box in the middle of google.com still takes a query. What changed is the furniture around it, and the reasoning behind the change is about how many questions you're supposed to ask in a row.

Heat: 1,790
An outlined app window in the foreground with a second window working behind it and a cursor path running between them
Product

Gemini Can Now Drive Your Mac, in a Window You Don't Have to Watch

Google's Mac app can now take a multistep job and finish it in apps it was never connected to, inside a window you don't have to watch. The part worth reading is where it stops and asks.

Heat: 1,740
A stack of library layers on a plinth with a locked version tag on top and streams feeding in from above
Product

llama.cpp Reaches v1.0.0 and Promises to Stop Breaking Your Build

llama.cpp reached v1.0.0 after three and a half years, and the release is about a promise rather than a feature. The promise is that the next breaking change to the C API gets a number before it breaks anybody's build.

Heat: 1,200
A blue app panel with an at sign, a circle and a small bar chart beside the words Gemini reads Wallet
Product

Gemini can read your Google Wallet spending, and it still can't move a cent

Dek: You can tag Google Wallet inside a Gemini chat and get a chart of last quarter's spending, and Gemini still can't move a dollar. The permission you grant is the part worth deciding on.

Heat: 1,400