Google has updated its Gemini Apps Privacy Hub so that chats are kept for up to two months even when a user has turned Gemini History off, up from 72 hours. Gizmodo reported the change on November 17, 2026. At the same time, Google says new controls let people keep their history while opting out of model training. The two changes pull in different directions, and the details are easy to misread.
AnthropicWhat changed in the Privacy Hub
Per Gizmodo, the Privacy Hub now says: "Even if Gemini History is turned off, future chats will still be saved for around 2 months." The archived page it cites, a snapshot dated November 12, 2026, listed a last update of October 13, 2026 and the 72-hour period. Gizmodo does not give an effective date, and notes that the Gemini History settings page still described a 72-hour period when it checked, expecting that to change.
Gemini History is the setting that saves conversations Gemini can remember and draw from. It defaults to 18 months. Users can change it to 3 months, 36 months or indefinite, turn it off, or delete saved chats at any time.
Two separate controls
Google pointed Gizmodo to a blog post titled "We're updating controls in the Gemini app." It says the new settings let people keep chat history while opting out of model training, and that users can still change how long history is saved and delete it manually.
| Control | What it governs | Reported behaviour |
|---|---|---|
| Gemini History | Whether Gemini remembers and draws on past chats, and for how long | Default 18 months; options 3, 18, 36 months or indefinite; can be turned off |
| Training opt-out | Whether chats are used to improve models | Now separable from history, per Google |
| Short-term retention | How long chats are kept even with history off | About 2 months, up from 72 hours |
The practical consequence is that "history off" no longer means "nothing stored beyond three days." Google did not comment to Gizmodo on why it extended retention, and said the update is meant to make Gemini safer, citing protection of "Google, its users, and the public." The notice notes human reviewers are involved.

How it compares
Gizmodo says the new window is about twice as long as competitors'. It reports that OpenAI holds temporary ChatGPT chats for up to 30 days and that Anthropic says Claude conversations are held for 30 days by default, with both companies saying they may keep some logs longer when necessary.
How long assistants say they keep chats
Stated retention in days, as reported by Gizmodo. Two months is drawn as about 60 days; companies may keep some logs longer when necessary.
Two caveats apply. The windows compare different features: Google's figure is for chats with history off, while OpenAI's is for temporary chats and Anthropic's is a default. And "up to" and "around" are not identical commitments, so the bars show stated figures, not guarantees.
What it means for users and teams
- Check which setting you changed. Turning off history and opting out of training are now different switches, so each needs checking.
- Treat "history off" as a shorter retention, not zero. For sensitive work, the stated window matters more than the toggle name.
- Regional terms may differ. Gizmodo does not say whether the change applies to particular regions, so organisations with data-residency duties should read the Privacy Hub for their location.
- Retention is a comparison point between assistants. Teams that use several assistants, including through Metir's multi-model chat, can use the stated windows as one input when deciding where sensitive prompts go.
Sources:
- Google to keep AI chat logs for two months even with Gemini history turned off - Gizmodo
- Everything That Happened in AI Today, Wednesday, November 18 - The Neuron
Image credits
Header and in-body image: Googleplex HQ, Mountain View, by Asoundd, Wikimedia Commons, CC BY-SA 4.0 (source). A file photo.