Gemini Apps Privacy Notice

Google Gemini
Last updated March 24, 2026 Privacy policy
+ Watch this document
8 Total
4 High severity
4 Medium severity
0 Low severity
Summary

This is Google Gemini's privacy policy explaining how your conversations with the Gemini AI assistant are collected, stored, and used. The most important thing to know is that Google keeps your Gemini conversations for up to 18 months by default, and human reviewers can read a sample of your chats to improve the AI — so anything you type into Gemini is not private. You can reduce data retention by going to myactivity.google.com and turning off Gemini Apps Activity, or deleting individual conversations from your history.

Technical Summary

This document is the Gemini Apps Privacy Notice, governing data collection, use, retention, and sharing practices for Google's Gemini family of AI products (Gemini app, Gemini Advanced, Gemini in Google Workspace, and related extensions), operating under Google's overarching Privacy Policy and supplemented by product-specific disclosures. The most significant obligations include Google's retention of Gemini conversation data for up to 18 months by default (3 years for certain Workspace and education accounts), human reviewer access to a sample of conversations for quality and safety purposes, and the use of conversation data to improve AI models unless users opt out. Notably, the policy explicitly warns users not to enter confidential information into Gemini, and that even after account deletion, conversation data may be retained for up to 18 months — a retention posture that deviates from minimization expectations under GDPR and creates heightened risk for users who inadvertently share sensitive data. The document engages GDPR (Arts. 5, 6, 13, 17), CCPA/CPRA, COPPA (given age-related disclosures for minors under 18), and the EU AI Act given Gemini's classification as a general-purpose AI system; material compliance considerations include the lawful basis for processing conversational AI data for model training, cross-border data transfer mechanisms for non-US users, and the adequacy of consent mechanisms for Workspace and educational account users whose data practices may differ from standard consumer accounts.

Institutional Analysis

REGULATORY EXPOSURE: This policy engages GDPR Arts. 5(1)(e) (storage limitation), 6(1)(f) (legitimate interests as processing basis for model training), 13 (transparency obligations), 17 (right to er…

REGULATORY EXPOSURE: This policy engages GDPR Arts. 5(1)(e) (storage limitation), 6(1)(f) (legitimate interests as processing basis for model training), 13 (transparency obligations), 17 (right to erasure — complicated by the 18-month retention window post-deletion), and 22 (automated decision-maki…

🔒

Compliance intelligence locked

Regulatory exposure, material risk, and due diligence action items.

Evidence Provenance
Captured March 30, 2026 06:16 UTC
Document ID CA-D-000326
Version ID CA-V-000388
Source URL https://support.google.com/gemini/answer/13594961
Wayback Machine View archived versions →
SHA-256 84e41d6d4eabc84744a0151c551f14547b75de066923dd2954e110c0ad31c776
✓ Snapshot stored ✓ Text extracted ✓ Change verified ✓ Cryptographically signed
Change Timeline
Analyzed Changes

5 changes analyzed since monitoring began.

Updated March 30, 2026

medium
What changed Google Gemini updated their Gemini Apps Privacy Notice on March 30, 2026. Change detected: 5 sentence(s) removed. Document contained 376 sentences after update.
Consumer impact Google Gemini removed the section of its privacy policy that told users they could opt in to letting Gemini personalize experiences using data from other connected Google apps. This makes it less clear what data Gemini uses for personalization and whether that feature has been removed or simply gone undisclosed. You can review your Gemini personalization settings in your Google Account to check what data connections are currently active.
Why it matters Users who relied on the opt-in personalization feature may no longer have clear information about how their data from other Google apps is being used within Gemini. The removal of this transparency language makes it harder to understand or control how Gemini personalizes experiences.

Updated March 28, 2026

medium
What changed Google Gemini updated their Gemini Apps Privacy Notice on March 28, 2026. Change detected: 4 sentence(s) modified. Document contained 381 sentences after update.
Consumer impact Google Gemini updated its privacy policy to explicitly include 'imported chats' as a category of data you share with the service and as part of your stored activity history. This means any chats you import into Gemini are collected and retained like other interaction data. You can visit Gemini Apps Activity to review and delete your activity, including imported chats.
Why it matters Users who import chats into Gemini should now know this data is explicitly collected and stored as part of their activity history. The update gives users clearer visibility into what data Google holds about them and how to delete it.

Updated March 27, 2026

high
What changed Google Gemini updated their Gemini Apps Privacy Notice on March 27, 2026. Change detected: 12 sentence(s) added. Document contained 381 sentences after update.
Consumer impact Google Gemini now explicitly states that data you import from other AI platforms — such as memories and chat histories — will be treated like regular Gemini activity, including being used to train Google's generative AI models. Additionally, if you opt in to 'Personal Intelligence,' data from connected Google apps will be used to personalize your Gemini experience. This significantly broadens the scope of data Google can use for AI training beyond what users might expect. You can manage or delete your imported activity at any time through your Gemini activity settings to limit how this data is used.
Why it matters Users who import chat histories or memories from other AI platforms into Gemini may not realize this data will be used to train Google's AI models. This expands the scope of data collection and use in ways that could affect privacy expectations and regulatory compliance.

Updated March 23, 2026

medium
What changed Google Gemini updated their Gemini Apps Privacy Notice on March 23, 2026. Change detected: 5 sentence(s) removed. Document contained 369 sentences after update.
Consumer impact Google Gemini removed disclosures explaining how it used data from connected Google apps to personalize your experience, leaving users without clear information about whether this data practice continues. The removal of this language means consumers can no longer easily understand the scope of data sharing between their Google apps and Gemini. You can review your Google account's connected apps and personalization settings at myaccount.google.com to check what data access remains active.
Why it matters Users who previously relied on this disclosure to understand how their Google app data was being used for Gemini personalization — and to opt in or out — no longer have that transparency. If the practice continues without disclosure, users lose meaningful control over a significant form of cross-app data use.

Updated March 21, 2026

low
What changed Google Gemini updated their Gemini Apps Privacy Notice on March 21, 2026. Change detected: 4 sentence(s) modified. Document contained 374 sentences after update.
Consumer impact Google Gemini's privacy notice now refers to its chat-based personalization feature as 'Memory,' making it easier to identify and locate the specific setting in the app. The substance of how your data is used has not changed — Gemini can still use sensitive information from your chats to personalize your experience when this feature is active. You can turn off Memory in your Gemini settings anytime to stop personalization based on your past chats.
Why it matters Formalizing the personalization feature as 'Memory' signals it is now a discrete, named setting users can find and control — which is relevant for anyone who wants to limit how their past Gemini conversations influence future AI responses.
High Severity — 4 provisions
Medium Severity — 4 provisions