
Google Extends Temporary Gemini Chat Retention From 72 Hours to Two Months
Google will hold off-history Gemini chats for about two months, up from 72 hours, while splitting chat history and AI training into separate privacy settings rolling out globally.
- By
- Tom Whitfield
- Filed
- Channel
- AI & Compute
- Read
- 4 min read
Google will keep temporary Gemini chats for roughly two months even when users turn chat history off, a twentyfold increase from the previous 72-hour retention window. The company says the longer holding period improves safety and security as its AI systems become more capable. The change rolls out globally in the Gemini app over the coming weeks.
The increase is confirmed by Google's own documentation. An archived version of the Gemini Apps Privacy Hub dated November 12 listed the old 72-hour retention figure. The updated policy replaces that limit with a window of around two months — a concrete shift in how long Google holds conversations that users explicitly tried to keep off the record.
How does Google's new retention period compare with rivals?
The new policy pushes Google past both of its main competitors in the conversational AI market. OpenAI states it may keep temporary chats in ChatGPT for up to 30 days. Anthropic retains incognito chats in Claude for 30 days by default.
All three companies carve out exceptions. They may hold some data longer than the stated windows for legal or safety reasons, or to investigate violations of their policies. Google's roughly 60-day baseline now stands at double the default retention of OpenAI and Anthropic.
The takeaway in plain words: users who disable history on Gemini now leave a longer data trail than on either competing chatbot.
What reasons did Google give?
Google did not answer questions about what prompted the change, and it did not clarify whether the shift affects how conversations are accessed or reviewed internally. The company instead pointed to a blog post.
That post gives three stated purposes for keeping chats longer: responding to users, processing feedback, and protecting Google's services and users. The updated Gemini Privacy Hub adds a fourth reason — potential system failures. It also contains one significant assurance: temporary conversations are not used to train Google's AI models.
The unanswered questions matter for commercial as well as privacy reasons. Enterprises evaluating Gemini deployments weigh retention periods against compliance obligations, and a two-month window with unexplained internal review practices gives corporate buyers less certainty than the vendor's rivals currently offer.
What changes for users and their settings?
The retention shift arrives alongside a structural change to Gemini's privacy controls. Google is introducing separate settings for chat history and AI training, breaking apart a configuration that previously bound the two together.
Until now, both functions hung on a single "Keep Activity" setting. Turning it off disabled both the saved chat history and the use of conversations for model training. Turning it on did the reverse.
Under the new scheme, the two controls work independently:
- Chat history: users can keep conversations in their history without contributing them to AI training.
- AI training: users decide separately whether Google may use their conversations to train its generative AI models.
The default states follow users' prior configuration. For anyone who previously disabled Keep Activity, both new settings arrive turned off. For users who had Keep Activity turned on, chat history remains enabled — but the training question becomes a separate, individual choice they must make.
In plain terms, the split gives users finer control over what Google saves, while the retention change means even the most privacy-conscious settings leave conversations stored far longer than before.
Why does the timing matter?
Google frames the longer retention as a response to AI systems growing more capable. As generative models handle more sensitive tasks and produce more consequential outputs, vendors face pressure to detect abuse, investigate incidents and defend their platforms — all activities that require data to examine after the fact.
The simultaneous settings split suggests a calculated trade. Users gain the ability to opt out of training while keeping their history, a long-requested configuration. In exchange, Google holds every conversation — including those flagged as temporary — for about two months.
For competitors, the move sets a new data point in the market. If two-month retention becomes the norm for safety-motivated data practices, OpenAI's 30-day temporary-chat window and Anthropic's 30-day default could come under pressure to lengthen as their models grow more capable too.
Google has not said whether the two-month window will grow again as Gemini's capabilities advance, but the company's stated rationale — safety in the face of more capable systems — points toward retention periods that scale with model capability rather than user preference.
Source: techcrunch.com
More from Tom Whitfield
Show full bio
Staff writer covering companies, deals and strategy in Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.
1,054 articles
Related articles
google-will-keep-gemini-chats-for-two-months-even-with-history-off
Google Will Keep Gemini Chats for Two Months Even With History Off
google-drops-paywall-on-gemini-task-agent-in-six-countries
Google Drops Paywall on Gemini Task Agent in Six Countries
