Whether Gemini uses your data to improve AI models depends on the product, account type, settings, and feature involved. Personal Gemini Apps and Google Workspace with Gemini should not be treated as the same environment.
Last verified: August 13, 2026. Product terms and controls can change. Confirm the current Google documentation and your administrator settings before relying on this guide.
The direct answer by environment
For a personal Google Account, Google says Gemini Apps activity can be used to improve Google services, including training generative AI models, when the applicable activity setting is on. Users can manage activity and related controls. Google also explains that some data may be retained for service and safety purposes even when activity is off.
For Google Workspace with Gemini, Google says customer Workspace data is not used to train or improve the generative AI and large language models that power Gemini and other systems outside Workspace without permission. Google's Workspace documentation also describes enterprise-grade controls and states that submissions in protected Workspace use are not used to train models or reviewed by humans.
Those are different commitments. A company policy should name the approved account and product rather than saying only “Gemini is approved.”
Personal Gemini Apps
Personal users should review Keep Activity, auto-delete, feedback, and any optional settings for audio, video, or screen-sharing data. Connected apps can add information from other Google services to the Gemini experience, which expands the data-handling question beyond the text typed into a chat.
For work, a personal account may be outside the organization's admin, retention, access, and support controls. Turning off one training-related setting does not automatically make the account an approved business environment.
Google Workspace with Gemini
Workspace use operates under organizational administration and Google Workspace terms. Existing Workspace permissions, data protection, and admin controls affect what Gemini can access and how the organization manages use.
Employees should still understand which Workspace account they are using, which Gemini surface is approved, what information is allowed, which connected apps are enabled, how generated content must be reviewed, and when to escalate.
An enterprise data commitment does not determine whether every document is appropriate for every workflow.
Gemini through Google Cloud or an embedded service
Gemini can also appear through Google Cloud, APIs, development tools, and third-party products. Those routes can have different contracts, configurations, logging, retention, and access paths.
Review the exact service. Do not transfer a statement about Gemini in Workspace to an API project, consumer app, browser extension, or third-party integration without checking the applicable terms and settings.
Training data is only one governance question
Leaders should also determine:
- which Google accounts and Gemini services are approved
- which Workspace content and data categories may be used
- whether connected apps are enabled
- who can create and share Gems or other configured experiences
- what human review is required
- how activity, retention, and deletion are configured
- what admins can audit or manage
- how employees report a mistake or uncertain use
These questions control daily behavior more directly than a general statement about model training.
What employees should check before using Gemini
Use a short pre-use checklist:
- Am I signed into the approved work account?
- Am I using a protected Workspace service or a personal Gemini environment?
- Is the information permitted in this product and workflow?
- Is Gemini connecting to Gmail, Drive, Docs, Calendar, or another service?
- Does the output need source verification, specialist review, or approval?
- Is there a required way to save, share, or delete the result?
Train employees with realistic examples. Ask them to classify public material, internal operations content, customer information, personal data, and regulated or privileged information. The answers should come from the organization's actual policy and Google configuration.
What administrators should document
Administrators and governance owners should maintain a product-specific record: enabled services, eligible users, account types, activity and retention settings, connected apps, data controls, sharing rules, audit capabilities, support owner, and last review date.
Recheck the record when Google changes a feature or terms, when a new connector is enabled, or when the organization expands use to a new data class or workflow.
Practical takeaway
Google distinguishes personal Gemini Apps from protected Google Workspace use. The safe enterprise answer is therefore product-specific: verify the exact account, service, settings, connected apps, and current terms.
Use the approved environment, minimize unnecessary data, preserve human accountability, and keep the product review current.