Google quietly revised its privacy policy on October 16, removing language that had promised content from Gmail, Drive and Photos would never be used to personalize advertising. Ars Technica reported the change on Friday, citing an earlier discovery by Android Authority. The update links the company's advertising systems to data that users may have shared with Google's AI products through a feature called Personal Intelligence.
Why it matters
For years, Google drew a clear line between the contents of its productivity and photo apps and the advertising business that generates most of its revenue. The old policy stated plainly that personalized ads would not be shown based on material stored in Drive, Gmail or Photos. That sentence is gone.
The replacement text is less absolute. According to Ars Technica, it creates room for data from those apps to feed ad personalization once it has been passed to other Google services. The publication notes that Google already operates a mechanism for exactly that kind of data sharing.
Privacy advocates are likely to treat the revision as a meaningful shift rather than a wording tweak, because it narrows a protection that users had been told was firm. The practical reach of the change depends on how many people have switched on the data-sharing feature at the center of it.
How Personal Intelligence connects the dots
Google launched Personal Intelligence in early 2026. The feature lets users connect data from several apps to Google's AI products, on the theory that more context produces better answers. Ars Technica reports that it links user data to the Gemini app as well as to AI Overviews and AI Mode in search.
Gemini itself does not currently carry ads. AI search results, however, can include sponsored content. The policy update appears to acknowledge that information routed through Personal Intelligence into AI search could therefore shape which ads a person sees.
Google disputes the framing that ads are built directly from private files. In a statement quoted by Ars Technica, the company said connecting apps is entirely opt-in and that ads remain tied to factors such as a user's query, the context of a response and their interests. The statement added that ads would not be based directly on a Gmail inbox or Photos library and that sensitive information would not be used to tailor advertising.
Ars Technica reads that response as an indirect confirmation that Workspace and Photos data can inform ads. The publication reasons that when someone uses AI Mode with Personal Intelligence active, summaries drawn from their data can feed the answers they receive — and because that same material is used for ad targeting, personal data becomes one ingredient in AI search advertising. A Google spokesperson told Ars that the clarification does not signal a change in how the company handles data with respect to advertising partners, that it is not scanning entire inboxes to build ad profiles, and that it does not hand this data to third parties.
Opt-in, but easy to overlook
Google is correct that Personal Intelligence requires a user to enable it. Ars Technica points out, however, that the company promotes the feature through prompts and pop-ups that may not spell out the consequences, meaning some users could have activated it without fully understanding what they agreed to.
Turning it off takes a few steps. The most direct route is through Gemini's settings for Personal Intelligence, where connected apps can be disconnected. The same controls sit several levels down in a Google account under Data & privacy, then Personalization settings, then Manage Personalized Recommendations in Search Services, then Connected Content Apps. From there, users can sever links to Workspace and Google Photos.
Ars Technica also draws a parallel to how Google handles AI interactions for model training. The company says personal data is not used to train its models, yet inputs and outputs are collected for that purpose. With Gemini wired into numerous services through Personal Intelligence, those outputs may carry fragments of personal information. Google has said it tries to screen such content out of user activity controls so it stays away from training data, but it does not guarantee success every time.
The broader tension, as Ars Technica describes it, is that Google treats cross-product data sharing as a competitive advantage for its AI work. Gmail retains protected status in the company's data handling, but the number of paths by which its contents can travel elsewhere has grown. Each additional connection a user makes across the Google ecosystem raises the exposure.
What to watch: whether regulators or privacy groups challenge the revised language, and whether Google adjusts its consent screens for Personal Intelligence. For now, the only sure way to keep Gmail and Photos content out of ad personalization is to check that the feature is switched off.
