TL;DR
- Ad Targeting: Google Search may use summaries and inferences from connected Gmail and Photos to personalize ads under an October 16 policy exception.
- User Choice: Connecting apps is optional; ad personalization and deletion of saved Search activity have separate controls.
- Personal Content: Search can train AI models on summaries containing a whole email when it is short or particularly relevant to a request.
- Gemini Chatbot: Gemini Apps chats are not used to show ads; Gemini’s connection settings operate separately from Search.
Google has added an opt-in exception to its privacy policy, allowing Search to use information drawn from Gmail and Google Photos to personalize ads. For people connecting those apps to get tailored AI answers, summaries and inferences from their personal content can also shape advertising.
The October 16 policy adds a qualification to Google’s exclusion of Drive, Gmail and Photos content from personalized ads: users can choose to make that data available to other Google services that support them. The October 1 version contained the exclusion without that exception. The permission depends on the policies and settings of the service receiving the information.
How Personal Answers Can Inform Ads
Personal Intelligence is Google’s feature for tailoring AI assistance with personal context, including information from connected apps. Its earlier addition to AI Mode in Search let answers draw on Gmail purchase and travel information and Photos context. AI Mode is Google’s conversational search experience.
Google’s Search help illustrates the benefit with a restaurant request for an upcoming trip. AI Mode can consult emails and photos for a hotel booking, flight itinerary and restaurants visited on earlier trips. It summarizes that information to recommend restaurants near the booked hotel. The user gets suggestions grounded in existing plans without having to supply every detail again.
Search can use those summaries, excerpts and inferences for ad personalization. Advertisers cannot access a connected Gmail inbox or Photos library, and Google excludes sensitive categories such as health, religion, race and sexual orientation from ad personalization.
Google described the policy language as a clarification in a response to Ars Technica. The company said the wording does not signal a change in its handling of data with advertising partners. It denied scanning entire inboxes to enhance advertising profiles or giving this connected personal data to third parties.
Google also told Search Engine Land it was starting an ad test for connected-app AI Mode users, according to October 19 reporting.
Connecting content apps grants access across Google’s Search services, including Search, Maps, Shopping and News, rather than just one AI Mode conversation. Search’s documented connections are for signed-in personal accounts aged 18 or older in supported regions. Its Workspace connection covers Gmail and Calendar; Calendar is limited to US English, while Photos connections are unavailable in some regions.
Personal Content Can Survive in a Summary
Google says Search trains its generative AI models on relevant summaries, excerpts, generated media and inferences, along with questions and responses, rather than directly on the underlying inbox or photo library. Before training, the company says it takes steps to reduce personal information, except when addressing abuse or harm.
A summary can retain the original personal content. Google’s guidance says a brief email or file, or one particularly relevant to the request, may itself become the summary.
A subset of connected information processed by AI Mode and Maps’ AI assistant, Ask Maps, is reviewed by people for debugging, safety and service improvement. This can include excerpts from emails, prompts and responses. Google says it takes steps to remove the account connection and reduce personal information before trained service providers review interactions, except when addressing abuse or harm. Reviewers generally receive derived Photos material; original library media can be reviewed when users submit feedback or in cases of abuse or harm.
Search and Gemini Have Different Rules
The Gemini chatbot also offers Personal Intelligence, but its advertising rule is different. The Gemini Privacy Hub, updated October 23, says Gemini Apps chats are not used to show ads. Changing Gemini’s connected-app settings does not change Search’s settings.
Gemini’s Keep Activity control governs future-chat use for model improvement. Turning it off stops that use unless the user submits feedback. Safety processing can continue: chats with Keep Activity off and temporary chats are retained for 72 hours, while chats already reviewed can be retained separately from the account for up to three years.
Ad Targeting, App Access and History Are Separate Choices
Users can turn off personalized ads through My Ad Center while keeping their content-app connections. Google then stops using saved account information to personalize ads, although contextual ads can still reflect an active search, page topic or location. Search can still use connected-app summaries for AI training and human review.
To stop Search’s connected-app access, users can disconnect Workspace or Photos in Search Services Personalization, under Additional Personal Intelligence and Connected Content Apps. AI Mode and Ask Maps remain usable without those connections.
Disconnecting an app or deleting the original email or photo leaves previously saved Search Services History in place. That retained activity can still be used for product improvement, including model training. Users manage and delete it separately through Google My Activity.
