
Share
A quiet privacy policy update means conversations users thought were fleeting will now linger for weeks. Here's what the change means for anyone who assumed "history off" meant "forgotten quickly."
If you've ever used a "temporary chat" feature and assumed your words vanished shortly after you closed the window, Google's latest policy update is worth your attention. The company is extending how long it retains Gemini conversations marked as temporary, or made with chat history turned off, from 72 hours to roughly two months. That's a nearly 20-fold increase in how long your supposedly private exchanges with an AI assistant sit on Google's servers.
Think of it like the difference between a note you scribble and toss in the trash versus one you leave in a locked drawer for weeks. Both feel private in the moment, but only one disappears quickly. Users who toggled off chat history in Gemini were, in effect, choosing the "toss it" option. Now that option looks a lot more like the drawer.
The change, detailed in Google's Gemini Apps Privacy Hub, will roll out globally over the coming weeks. An archived version of that support page from November 12 confirms the prior 72-hour window, making the shift easy to document even though Google hasn't publicized it with much fanfare. Google says the extended retention is meant to improve safety and security as its AI systems grow more capable, and its blog post on the matter points to responding to users, processing feedback, and protecting its services as the underlying rationale. The updated privacy page also cites potential system failures as a reason to hold onto data longer, and it states clearly that temporary conversations are not used to train AI models.
That last point matters for anyone trying to parse what's actually changing here. Retention and training are two separate questions, and Google is answering only one of them with this update. Your temporary chats won't feed the next version of Gemini. But they will exist, accessible to Google, for about two months instead of three days.
Context helps here, because Google isn't acting alone in this space, and it isn't even the most aggressive among its peers. OpenAI says it may keep temporary chats in ChatGPT for up to 30 days. Anthropic retains incognito chats in Claude for 30 days by default. Google's new two-month window actually exceeds both of those competitors' default retention periods, even though its prior 72-hour standard had been notably shorter.
All three companies include an important caveat: they may hold onto data longer than their stated defaults for legal reasons, safety investigations, or to look into potential policy violations. That caveat is the kind of fine print that rarely gets attention until it matters, say, during a legal dispute or a safety review, at which point users may discover that "temporary" was always a flexible term.

Google did not answer direct questions about what prompted the longer retention period or whether it would change how conversations get accessed or reviewed internally. That's a notable gap. When a company extends how long it holds onto user conversations by a factor of twenty, the public reasonably wants to know who looks at that data, under what circumstances, and with what oversight. Pointing to a blog post that offers general justifications isn't the same as answering those specific questions.
The timing of this change coincides with another, arguably more consumer-facing shift. Google is separating the settings that control chat history from the settings that control AI training. Previously, both lived under a single toggle called "Keep Activity." Now users get two distinct choices: whether to keep their chat history, and separately, whether Google can use their conversations to train its generative AI models. It's a reasonable move toward more granular consent, letting people keep a record of their own conversations without automatically handing that content over for model training.
Google says both new settings will default to off for users who had previously disabled Keep Activity, preserving the spirit of their original choice even as the underlying system changes. Users who had Keep Activity turned on will keep their chat history enabled going forward, but they'll now get a separate, explicit choice about whether their conversations feed AI training. That's a meaningful improvement in transparency, even if it arrives bundled with a far less transparent extension of how long temporary data sticks around.
Data retention policies rarely make headlines the way product launches or AI capability breakthroughs do, but they shape the actual risk profile of using these tools every single day. A two-month retention window means more time for that data to be subject to breaches, legal subpoenas, internal access, or simple human error. It also means more time for policy to quietly shift again, as it just did here, without most users ever noticing.
The deeper issue is one of trust and expectation. When someone toggles off chat history or opens a temporary chat, they're making an informed guess about what that choice means for their privacy. If the underlying rules change substantially, and the company doesn't clearly explain why, that guess becomes unreliable. Multiply that uncertainty across the hundreds of millions of people using Gemini, ChatGPT, and Claude, and you get a landscape where basic privacy controls carry less certainty than users assume.
None of this means Google's reasoning is unfounded. Safety reviews, fraud investigation, and system reliability are legitimate concerns for any company running AI services at scale. But legitimate reasons for data retention still deserve clear, specific public explanation, not a quiet update to a support page that most users will never read. As AI assistants become woven into more of daily life, from drafting emails to answering medical questions, the gap between what people think they're agreeing to and what companies actually do with their data is a gap worth closing, not widening.
Tags
Original Sources
Google to keep Gemini chats for 2 months, up from 72 hours, even with history off | TechCrunch
↗ https://techcrunch.com/2026/11/17/google-to-keep-gemini-chats-for-two-months-up-from-72-hours-even-with-history-off
About the author
Amara's entry point into AI was an epidemiology role at a London research hospital, where she spent five years studying how digital health tools reached — or conspicuously failed to reach — underserved communities. Watching early algorithmic systems in healthcare quietly entrench existing inequalities, she redirected her career toward the systemic consequences of AI at scale. She covers AI through an unflinching lens: who benefits, who bears the cost, and what evidence actually says versus what the press release claims. Her writing is calm and precise, but she doesn't mistake balance for neutrality.
More from The Steward →This Week's Edition
18 November 2026
29 articles
Related Articles

Who Decides If Your Treatment Is Medically Necessary? Industry Leaders Say It Shouldn't Be AI Alone
Policy & Regulation · 6 min

Epic Systems Pushes Back on Antitrust Criticism, Pledges Broader AI Data Access
Policy & Regulation · 5 min

AI Is Showing Up in Court Filings. Courts Can't Yet Agree on How to Prove It
Policy & Regulation · 6 min
Related Articles

Who Decides If Your Treatment Is Medically Necessary? Industry Leaders Say It Shouldn't Be AI Alone
Policy & Regulation · 6 min

Epic Systems Pushes Back on Antitrust Criticism, Pledges Broader AI Data Access
Policy & Regulation · 5 min

AI Is Showing Up in Court Filings. Courts Can't Yet Agree on How to Prove It
Policy & Regulation · 6 min
More Stories
© 2026 Cedar & Bloom. All rights reserved.