Ahead of Q3, Google redraws another line between personal context and ads

Alphabet heads into its third-quarter results this week with a problem most companies would quite happily borrow: Google is already almost everywhere.

AI Mode has surpassed one billion monthly active users, while AI Overviews reaches more than 2.5 billion every month. Gmail has three billion users. The Gemini app crossed the one-billion monthly-user mark in August, becoming Google’s 14th product to reach a billion users. In May, Google said five of its products had more than three billion users each. 

When products come with commas like those, finding the next advertising opportunity becomes less about finding another audience and more about understanding the audience already there a little better.

Read On: Google turns 28: Searching for answers in a world it helped create

That makes the timing of Google’s latest advertising experiment worth watching.

Alphabet is due to report its Q3 2026 results on October 28. In Q2, Google advertising generated $81.63 billion, including $63.27 billion from Google Search & other and $11.06 billion from YouTube ads. Search & other revenue grew 17% year-on-year. Alphabet said its AI features were driving Search query growth and highlighted improvements in ads quality, advertiser tools and AI user experiences, with retail and finance making the largest contributions to Search & other revenue growth. 

And less than two weeks before the next set of numbers arrives, Google began testing ads in one of the more intimate corners of its increasingly AI-shaped Search business: Personal Intelligence.

Search, now with a backstory

Personal Intelligence allows users to connect services including Gmail and Google Photos to AI Mode so Google can answer questions using information about their own lives.

You've probably used it yourself, and have been mildly touched at how AI mode remembers your dietary preferences and recent ailments, even as it uses previous restaurant reservations to make a new restaurant recommendation. 

You're also being mildly touched in many other ways. 

Read On: #e4mXplains: New model, who dis? Inside Big Tech’s AI marketing arms race

Personal Intelligence can connect information scattered across emails, photographs and other Google services to understand not simply what someone is asking, but the circumstances behind the question. 

The feature expanded this year to nearly 200 countries and territories across 98 languages, after launching in India in April.

On October 16, Google began testing ads in AI Mode for users who have chosen to connect those apps.

Advertisers do not receive that information.

Google says ads will not be based directly on a user’s Gmail inbox or Google Photos library, advertisers cannot access those connected apps, and sensitive categories such as health, religion or sexual orientation remain excluded from personalised advertising.

But Search may use “summaries, excerpts and inferences” generated from those sources to personalise an experience. And Google now says that same information may also be used to show ads.

Nike, in other words, does not get to know about your flight to Srinagar.  

Google may know about it, understand why you suddenly need a sports coat, and decide that Nike has something useful to say.

Read On: YouTube wants a share of India's shopping carts. Will ad budgets follow?

For Saurabh Kumar, Founder of performance marketing agency Envigo, “the shift is in context, not in the targeting tools.”

The controls available to advertisers remain broadly familiar, he argues. What changes is the selection logic behind them: a search can increasingly be understood alongside what someone has bought, booked or planned.

Mihir Mehta, Co-Founder and Managing Partner at 0101.Today, describes the resulting asymmetry bluntly: “The advertiser may know less about the individual, while the platform knows more.”

And that is where things become more interesting than another targeting upgrade, and so consequential for advertisers across the board. 

‘Never’ acquires an exception

Google has not always described the boundary this way.

In previous privacy messaging, the company was strikingly categorical about information stored in some of its most personal products.

One Google privacy FAQ said content in Drive, Gmail and Photos was “never used for ads purposes.” That same page now carries an October 16, 2026 update explaining that users can choose to connect Photos and Workspace apps to other Google services, and that depending on their settings, a subset of that data can also be used to show ads.

Read On: Beyond Reach: Can AI-generated ads turn attention into action?

Google CEO Sundar Pichai had gone further in earlier privacy messaging, saying Google did not use information in apps where people primarily store personal content (including Gmail, Drive, Calendar and Photos) for advertising purposes, “period.”

The word “period” has, effectively, acquired a comma.

That does not mean Google has suddenly opened Gmail inboxes to advertisers. It has not.

The new arrangement is opt-in. Users choose whether to connect content apps, can disconnect them later, and advertisers still cannot see the underlying Gmail inbox or Photos library. Google's current help documentation explicitly preserves those safeguards.

What has changed is what Google itself can do with information derived from those services once a user chooses to connect them.

The old boundary was straightforward: personal content stored here is not used for advertising.

The new one is more precise: ads are not based directly on that content, but summaries, excerpts and inferences generated from it may help determine what appears.

For Kumar, the entire transparency question therefore “sits in the word ‘directly’.”

An inference can carry highly personal context even if the original email or photograph remains private, he argues. The concern begins when somebody opts in because they want a better AI assistant without necessarily appreciating that the same understanding can also make advertising more personalised.

Except Google has always inferred things

Rahul Vengalil, CEO and Co-Founder of tgthr, provides a useful objection to treating this as the invention of a new advertising paradigm.

Google has spent years combining signals across Search, YouTube, location, demographics, affinity categories and in-market behaviour to build useful pictures of consumers, he points out.

Read On: Who's really in control? How Meta, Google and Amazon are rewriting the rules of AI advtg

Personal Intelligence therefore adds more information to an already sophisticated system rather than inventing the practice of inference itself.

Nor, Vengalil argues, is Google withholding underlying personal information from advertisers a new safeguard. Advertisers have long acted on Google's audience and intent signals without receiving the personally identifying information from which those signals were constructed.

He expects the additional context to produce perhaps a 5-10% incremental improvement in advertising effectiveness, rather than anything approaching a 50% transformation. That is his estimate rather than campaign data, but the larger point is clear: to the person buying media, Personal Intelligence may be evolution rather than revolution.

At Google's scale, of course, “incremental” is rarely an uninteresting word.

The company has not publicly disclosed how many of AI Mode's billion-plus users have chosen to connect personal-content apps, so the total Personal Intelligence advertising audience cannot simply be equated with AI Mode's overall reach.

But neither is the experiment buried in some forgotten Google beta.

Consent becomes the awkward bit

Venkat Mallik, Founder and CEO of J7, sees the change less as a targeting revolution than as a shift in the terms users are agreeing to.

Personal Intelligence was initially an ad-free experience for connected users. Advertising has now entered it.

Read On: What does an ad look like inside AI? ChatGPT and Google are starting to answer

Mallik argues that the shift is visible not only in Google’s broader privacy language, but in Personal Intelligence itself. Google had earlier said users connecting apps through Personal Intelligence would not see ads, he notes as well. “This has now been reversed and to that extent it’s a change in terms.”

For him, the more important question is whether consent travels with the product as its purpose expands. “Consent given for one purpose, a smart AI assistant, doesn’t automatically extend to advertising.”

Google's current system is explicitly opt-in, and users retain controls over which apps are connected.

The more difficult question is whether permission remains intuitively meaningful when one system performs several jobs at once.

A user may understand perfectly well that connecting Gmail makes an AI assistant better at remembering restaurant reservations, travel plans or purchases.

Whether that user also understands that insights produced from the same connection may influence advertising is another question.

Rishabh Mahendru, Senior Vice President, Client Success and Growth at AdLift, reduces the issue to a useful test: “If an ad appears because Google inferred something from your email or photos, how would you know?”