Quick Answer
Gemini now keeps future chats for around two months when History is off, rather than the prior 72-hour period. Google says those chats are not used to train its AI models unless you submit feedback, but they can be reviewed for service and safety purposes. Review Gemini’s separate History and training controls before sharing sensitive information.
Key Takeaways
- Google split Gemini chat History and generative-AI training into separate controls on November 16, 2026.
- Turning History off no longer means Gemini deletes future chats after 72 hours.
- History-off chats remain stored for around two months, according to Google.
- Temporary chats and History-off chats are not used for AI training unless feedback is submitted.
- Human reviewers and service providers may review some chats for safety, service operation, and feedback handling.
What changed with Gemini chat retention?
Google changed Gemini chat retention by separating the former combined control into one setting for chat History and another for generative-AI model training. The company announced the change on November 16, 2026, and said users can keep normal chat history without agreeing to have that chat content used to train its generative AI models. Google’s announcement describes the new controls as independent choices.
Gemini’s revised approach gives users more control over two decisions that previously moved together. Keeping a searchable record of useful conversations can be convenient when Gemini has helped with research, planning, or writing. Opting out of model training addresses a separate privacy concern: whether the content of those conversations contributes to future AI development.
The important limitation is that an AI-training opt-out does not mean Google immediately deletes every chat. Gemini still needs to retain some content for the service to function, process user feedback, and address abuse or safety concerns. Users should therefore treat the History control and the training control as related but distinct privacy choices.
Does turning Gemini History off stop AI training?
Turning Gemini History off stops future History-off chats from being used to train Google’s AI models, unless the user chooses to submit feedback on a chat. Google’s Gemini Apps Privacy Hub says temporary chats and chats created with History off are not used for model training under the updated policy. The exception matters because feedback can include a conversation for evaluation and improvement purposes.
Gemini’s separate controls are a meaningful improvement for people who want to retain ordinary conversations without using them for model training. The new control structure allows a user to keep chat history while independently declining AI-training use, rather than forcing an all-or-nothing decision.
Users should still avoid entering information that would cause serious harm if exposed or reviewed, including account passwords, Social Security numbers, full payment-card data, private medical records, and confidential work documents. A training opt-out limits one category of use. It does not turn Gemini into an offline, end-to-end encrypted personal vault.
Why does Gemini keep History-off chats for around two months?
Gemini keeps History-off chats for around two months because Google says it needs the conversations to respond to users, process feedback, and protect Google, its users, and the public. That retention period applies to future chats created after History is turned off. Google’s policy describes the retention as “around two months,” which means users should not interpret the setting as an immediate deletion command.
The two-month window is longer than the prior 72-hour retention period associated with these chats. The retention change is part of the same rollout that separated chat-history management from training permission. The practical consequence is simple: turning off History reduces long-term account history, but it does not prevent short-term service retention.
Gemini chats reviewed by service providers are handled under a separate, more consequential rule. Google says those chats are separated from the user’s account and may be retained for up to three years. Separation from an account reduces direct account association, but it does not make sensitive content safe to share. Avoid entering details that identify you or another person unless the task genuinely requires them.
How does the new Gemini retention period compare with the old one?
Gemini’s new History-off retention period is around two months, replacing the previously cited 72-hour period for those chats. The change gives Google more time to operate the service and handle feedback or safety issues, but it also means users who selected History off under the old expectation should reassess what they share.
| Gemini chat choice | Training use | Retention described by Google | Practical meaning |
|---|---|---|---|
| History on, training allowed | Chat content can be used for generative-AI model training | Chat history remains available in the account | Useful for retrieving past work, with the broadest data-use permission |
| History on, training off | Chat content is not used for AI training | Chat history remains available in the account | Best fit for users who want searchable history without training use |
| History off | Not used for training unless feedback is submitted | Around two months | Reduces account history but does not provide immediate deletion |
| Temporary chat | Not used for training unless feedback is submitted | Subject to Google’s temporary-chat handling | Useful for one-off prompts, but not a substitute for careful data sharing |
Gemini’s retention distinctions are especially relevant for users who use AI tools for financial planning, shopping, or account-related questions. For example, Gemini can access spending information in some Google services through features such as Google Wallet finance insights. Financial prompts should contain only the minimum detail needed to receive useful guidance.
What do the separate Gemini controls mean for privacy?
Gemini’s separate controls mean users can make a more precise privacy choice: keep chats for personal reference, decline model-training use, or turn off History for future chats. Google’s previous combined control made chat retention and AI-training permission harder to distinguish. The revised design gives users a clearer way to match settings to their intended use.
Gemini privacy controls still have limits because service operation is broader than AI model training. Google says History-off chats can be retained to respond to users, process feedback, and protect the company, users, and the public. Human reviewers may also help with those functions, which is why sensitive information should remain outside ordinary chatbot prompts.
Google’s privacy design differs from a temporary-thread feature that keeps searches out of a visible account record. The distinction matters because temporary AI search threads and Gemini conversations can have different product rules and retention practices. Check the privacy terms for the specific Google product you are using rather than assuming one setting applies across every AI feature.
What should you avoid sharing in Gemini chats?
Gemini users should avoid sharing high-risk personal, financial, health, and workplace information because History-off chats can remain available to Google for around two months and may be reviewed for service or safety purposes. A prompt that includes a password, account recovery code, tax document, or private client file creates unnecessary exposure without improving most AI responses.
- Account passwords, one-time codes, recovery keys, and authentication backups.
- Full bank-account numbers, payment-card numbers, and government identification numbers.
- Unredacted medical records, insurance documents, and highly personal health details.
- Confidential business plans, source code, contracts, and customer data.
- Private information about another person who has not agreed to its use.
Gemini can still help with many sensitive-adjacent tasks when users remove identifying details. Replace a real name with a placeholder, summarize a document instead of pasting it in full, and ask for general explanations rather than individualized financial, legal, or medical decisions. Users seeking help after a data breach should also prioritize official account-security steps, as incidents involving exposed AI chat transcripts show why chat content deserves careful handling.
How should you review Gemini’s new privacy settings?
Gemini users should review both the History setting and the generative-AI training setting because the controls now govern different outcomes. Check whether you want past chats to remain available in your account, then separately decide whether Google can use eligible chat content for model training. Google’s policy change means one setting no longer answers both questions.
- Open Gemini’s privacy or activity controls while signed in to the Google account you use for Gemini.
- Review the History setting and decide whether future chats should appear in your account history.
- Review the separate generative-AI training setting and turn off training use if you do not want eligible chats used for that purpose.
- Use temporary chats for one-off conversations when you do not need them in normal chat history.
- Delete any existing sensitive chats that remain visible in your account, while recognizing that deletion does not override every retention rule described in Google’s policy.
Gemini users should stop and contact Google support if an account shows activity they did not create or if privacy controls appear unavailable, because an account-access problem requires security review rather than routine preference changes. Google’s settings can limit future data use, but the most reliable protection remains entering less sensitive information from the start. Privacy-setting guidance also emphasizes reviewing the updated controls as the rollout reaches users.
FAQ
Does Gemini delete chats immediately when History is off?
Gemini does not delete future History-off chats immediately. Google says those chats are retained for around two months to respond to users, process feedback, and protect Google, its users, and the public.
Can Google train Gemini on History-off chats?
Gemini does not use History-off chats to train Google’s AI models unless the user submits feedback. Feedback can create a separate review path, so users should avoid including sensitive details even when History is off.
Can I keep Gemini chat history without AI training?
Gemini now allows users to keep chat history while separately opting out of generative-AI model training. Google introduced the split controls on November 16, 2026.
How long can reviewed Gemini chats be kept?
Gemini chats reviewed by service providers may be retained for up to three years after being separated from the user’s account. Google says this handling supports service operation, feedback processing, and safety protections.
Are Gemini temporary chats private?
Gemini temporary chats are not used to train Google’s AI models unless feedback is submitted, but they are not equivalent to an offline private notebook. Users should still keep passwords, financial details, private health records, and confidential work information out of temporary prompts.
