TLDR;
Google is testing ads in AI Mode for people who choose to connect apps through Personal Intelligence. The platform can use limited insights from those apps when deciding which ad to show, but advertisers cannot see a person's Gmail or Google Photos. For brands, this creates a potentially useful new signal without new targeting controls, so it should be treated as a measured test rather than a precise audience product.
What happened
Google confirmed on 16 October that it has started testing ads in AI Mode for people who opt into Personal Intelligence. Search Engine Land reports that these ads can reflect the query, the context of the AI-generated answer, the user's existing interests and limited insights from connected apps. Google has not announced new advertiser controls, targeting options or a wider rollout.
Personal Intelligence lets a signed-in user connect Google content apps to Search. Google's support documentation says AI Mode can use summaries, excerpts and inferences from Gmail, Calendar and Google Photos to personalise responses. That could include past purchases, reservations, preferred brands or dining preferences. The same processed information may now help Google select a more relevant ad.
The boundary matters. Google says ads will not be based directly on the contents of a Gmail inbox or Google Photos library. Advertisers cannot access those connected apps, and sensitive categories such as health, religion, race or sexual orientation are excluded from personalised advertising. Users must choose to connect the apps and can disconnect them later.
This is a change from the earlier Personal Intelligence experience in AI Mode, which launched without ads. Google's March expansion made the feature available to US personal accounts and described tailored shopping and travel recommendations based on connected context. An updated Gmail explanation and the current privacy policy now make clear that, depending on settings, a subset of connected data can also support personalised advertising.
Why it matters
AI Mode already changes the shape of a commercial query. A person can describe a need, add constraints and refine the answer over several turns. Personal Intelligence adds private context that the shopper may never type into the prompt. An ad can therefore appear relevant because Google understands the surrounding task, even though the advertiser cannot inspect the signal that produced the placement.
That creates an attribution problem as well as an opportunity. Campaign teams may see an impression and a conversion, but they will not know whether the decisive factor was the query, the generated answer, an existing interest or a connected-app inference. Conventional audience labels will not explain why two people asking a similar question see different advertising.
The privacy line also has to be communicated accurately. Saying that advertisers can target Gmail or read connected apps would be wrong. Saying connected information has no role would also be wrong under the new test. Enterprise legal, media and customer teams need one agreed description: Google may use processed connected-app insights inside its own systems, while the advertiser receives no access to the underlying content.
There is no evidence yet that the extra context improves incremental sales, customer quality or lifetime value. Google has not supplied results, eligible campaign types or a rollout schedule. The commercial case therefore depends on controlled experiments and first-party outcomes, not the novelty of reaching a highly personalised AI session.
How your brand can benefit / be affected
Check whether your Google Ads campaigns can appear in the test and which markets, accounts and formats are eligible. Do not invent a Personal Intelligence audience segment or promise targeting precision that the advertiser cannot verify.
Build creative around the task rather than the assumed private signal. Product claims, prices, availability and landing-page promises should work for anyone who sees the ad. If relevance depends on knowing a hidden personal detail, the message is too fragile and risks feeling intrusive when Google's inference is wrong.
Measure downstream quality with consented first-party data. Compare assisted and last-click conversions, profit, cancellations, returns and customer-service contacts with similar AI Mode traffic. Hold back spend where possible, annotate the launch date and separate any uplift from broader changes to bids, creative or product availability.
Give privacy and governance teams the Google documentation. Review consent, regional requirements, retention and the distinction between platform personalisation and advertiser data access. The benefit will come from accurate offers and disciplined measurement, not from pretending a black-box contextual signal is a customer list.
News date: 16 October 2026. Editorial review: 17 October 2026.