Google has added a conditional exception to its pledge never to personalize ads from Gmail, Drive, or Photos content, and it applies only to people who connect those services to Search. The revised privacy policy took effect Oct. 16, 2026. The Verge reported the change on Oct. 23, and PPC Land published a line-by-line analysis of the text the same day.
The pledge dates to July 1, 2021. Google’s new wording keeps it but attaches a condition: no personalized ads from that content “unless you choose to make data from these services available to other Google services that support personalized ads.” PPC Land compared the new version with 54 earlier ones in Google’s policy archive and found no previous carve-out of this kind.
The mechanism is Personal Intelligence, an opt-in feature that lets AI Mode, Google’s conversational search experience, read a user’s own email, photos, and files. The policy says Search may process data a user allows it to reach “to personalize content, including ads.” Per PPC Land, the policy also lists personalized content as a third basis for ad personalization, next to interests and activity.
The scope is tighter than the headline suggests. According to Google’s help page, as PPC Land describes it, connecting is open to adults using a personal account. Connectable apps are currently Gmail, Calendar, and Google Photos, and Photos is not offered in every region. The policy text also names Drive, a difference from the help page list.
Advertisers gain no window into private content. The help page says ads will not rest “directly” on an inbox or photo library, and that advertisers cannot access either. What can feed ad selection is derived material: summaries, excerpts, and inferences. The same page concedes that a summary of a short email may be the email itself, which blurs the line between direct and indirect.
Two protections stay intact. The ban on personalized ads built on sensitive categories (race, religion, sexual orientation, health) is unchanged, as is the rule against sharing personally identifying information with advertisers. Neither the policy nor the help page describes how an email about a doctor visit is treated differently from one about a flight, even though AI Mode may summarize both.
Gemini plays a smaller role than the wording implies. The permission sentence names Search and Gemini Apps as services a user may allow to read other Google data, but only Search gets the ad-processing description. Google VP Dan Taylor said in January 2026 that the Gemini app carries no ads and that no change is planned.
For search marketers, the effect runs through AI Mode monetization. Google confirmed around Oct. 16 that it had begun an ad test in AI Mode for people who have connected apps, though in March it had said those users saw none. A person asking about a hotel reservation buried in an email expresses intent the query text alone would not reveal. Under the new policy, the inference behind the answer can also shape which ad appears.
None of that becomes a buyable audience. PPC Land found no new targeting controls, segments, or reporting tied to this data. Advertisers cannot see whether a connected-app inference helped select an AI Mode ad, and Google has published no count of users who have connected apps. Any lift would arrive inside Google’s own ad selection.
Trust is the larger variable. Ad personalization stays under My Ad Center, while connecting apps is a different setting, so someone who connects Gmail to get better answers while ad personalization stays on is covered by the exception. The policy describes no extra consent step for the ad use.
Opt-in rates are unknown. Teams buying on Google should read AI Mode results with that blind spot in mind, and brands that rely on user goodwill should watch how many people choose to connect.
Based on reporting by PPC Land, published Oct. 23, 2026, which analysed Google’s privacy policy effective Oct. 16, 2026 and cited The Verge’s report of the change.
