Google has started testing ads in AI Mode for the one group of users it previously kept ad-free: people who connect Gmail and Google Photos through Personal Intelligence. Google confirmed the test to Search Engine Land on October 16, 2026 — and the detail that matters is buried in a single word of its updated help page: “directly.”
This is a limited test, not a rollout, and Google has not disclosed its size or which regions are included. But the direction is clear, and it changes what advertisers and SEOs should expect from AI Mode next.
Key Takeaways
- Google confirmed on October 16, 2026 that it is testing ads in AI Mode for users who connect apps via Personal Intelligence, reversing its March position.
- Ads will not be based “directly” on your Gmail inbox or Photos library — but summaries, excerpts, and inferences from them may be used to show you ads.
- Connecting apps is opt-in and can be undone at any time; personalized ads stay off for sensitive categories like race, religion, sexual orientation, and health.
- Advertisers get no access to anyone’s inbox or photos and no new targeting controls — the signals change which ads some users see, not what advertisers can buy.
- This is a limited test that may change; treat every detail as provisional.
What Did Google Actually Announce?
Google confirmed to Search Engine Land on October 16, 2026 that it is beginning to test ads in AI Mode in Search for users who connect their apps through Personal Intelligence. The same connected-app information that personalizes AI Mode’s answers — insights and summaries drawn from Gmail, Google Photos, and other linked apps — may now also personalize the ads those users see.
Ads in AI Mode are not new. What is new is the audience: until now, the Personal Intelligence experience was ad-free. Google says the ads will still reflect the user’s query, the context of the AI-generated response, and the user’s interests — the connected-app layer sits on top of those signals. Google has also updated its privacy policy to explain how connected-app information may be used for ad personalization. It has not announced a wider rollout, new advertiser controls, targeting options, or reporting for the test.
What Does “Not Directly” Mean in Google’s Help Page?
Google’s updated help page states: “Ads will not be based directly on your Gmail inbox or your Google Photos library.” The next sentences do the real work: “If you connect your content apps to Search Services, Search may reference summaries, excerpts and inferences from these sources to personalize your experience and do things for you. That same info may be used to show you ads, so they are even more helpful for you.”
Read those two statements together and the distinction is precise. An ad is not matched to the contents of a specific email or photo. But what Search concludes from your emails and photos — the summaries, excerpts, and inferences — can shape which ads appear. The word “directly” is doing the heavy lifting: it separates the raw data, which stays out of ad targeting, from the derived understanding, which does not. Marketers should quote this wording exactly when briefing clients, because paraphrases like “Google reads your email to show ads” and “your data is never used for ads” are both wrong.
Why Is This Test a Reversal of Google’s March Position?
In March 2026, when Google expanded Personal Intelligence in AI Mode, a Google spokesperson told Search Engine Land: “There are currently no ads for people who choose to connect their apps with AI Mode. That isn’t changing right now.” The same statement did leave the door open — “In the future, we anticipate that ads will operate similarly for people who choose to connect their apps with AI Mode” — but the present-tense promise was ad-free.
Seven months later, the future arrived. This matters less as a gotcha than as a planning signal: features that launch ad-free inside Google’s AI surfaces should be treated as pre-monetization, not permanently ad-free. If your strategy assumed Personal Intelligence users were unreachable by ads, that assumption expired on October 16. The protections Google launched with — opt-in connection, disconnection at any time, no advertiser access — remain in place; the ads are what changed.
What Protections Has Google Kept in Place?
Google lists four protections around the test, and they are worth stating exactly because they define what this is and is not:
- Opt-in only. Users must choose to connect their apps to Search Services, and can disconnect them at any time.
- No direct targeting. Ads will not be based directly on the contents of a Gmail inbox or Photos library — only on summaries, excerpts, and inferences.
- No advertiser access. Advertisers cannot access users’ connected apps, and Google has announced no new controls or targeting options for this test.
- Sensitive categories excluded. As Google’s help page puts it, “Google doesn’t show you personalized ads based on sensitive categories, such as race, religion, sexual orientation, or health.”
What Google has not said: how many users are in the test, which countries are included, or whether connected-app signals will personalize ads on other surfaces. Anyone claiming to know those answers is guessing.
How Does Google’s Approach Compare With OpenAI’s ChatGPT Ads?
The Keyword frames this as a Google-versus-OpenAI contest over personal context. OpenAI is moving the same direction from the opposite starting point: its monetization lead, Asad Awan, has described memory-based ChatGPT ads chosen partly by what ChatGPT remembers about a user, and OpenAI’s help article says its system considers “the context and intent of the current conversation” and, when ads personalization is enabled, “select signals from a user’s broader ChatGPT experience.”
The difference is where the signal lives:
| Google AI Mode | ChatGPT | |
|---|---|---|
| Signal beyond the query | Summaries, excerpts, and inferences from connected Gmail and Photos | Signals from the user’s broader ChatGPT experience, including memory |
| When it applies | When a user connects Google apps to Search Services | When ads personalization is enabled |
| Stated limits | No personalized ads on sensitive categories; no advertiser access to Gmail or Photos | Ads kept separate from answers; some conversations should never carry an ad, per OpenAI |
OpenAI draws on a user’s history with ChatGPT itself. Google draws on email and photo accounts many people have used for years — a depth of personal context no chatbot holds. That asymmetry is why this test, however small, is the more significant of the two stories. For how ChatGPT’s own ad program works in practice, see our ChatGPT ads setup guide and the original ChatGPT ads explainer.
What Should Advertisers and SEOs Do About It?
There is nothing to buy yet — no new campaign type, control, or report — so the practical response is preparation, not panic:
- Brief clients with the exact wording. The story will be misreported as “Google targets ads from your emails.” The help page’s “directly” distinction is your correction.
- Expect AI Mode ad inventory to keep expanding. Combined with Google moving links higher in AI Mode and AI Overviews — covered in our AI Mode links update — the pattern is a surface being built for both organic visibility and paid placement at once.
- Watch your own connected accounts. If you use Personal Intelligence, you are in the population this test targets. Check your connection settings and know what you opted into.
- Don’t reallocate budget on a test. Google has not disclosed scale, regions, or a rollout timeline. Note it, monitor it, and wait for announced controls before planning spend around it.
Final Thoughts
The honest summary is narrow: Google is testing ads for a previously ad-free slice of AI Mode, personalized by inferences from apps users chose to connect, with advertisers gaining no new levers. The significance is directional. Personal context is becoming an ad signal on both major AI assistants, the word “directly” is where the privacy line now sits, and the March promise that connected-app users would stay ad-free has been retired. It is a limited test and it may change — but the direction it points is unlikely to reverse.
People Also Ask
Are ads in AI Mode based on my Gmail emails?
Not directly, per Google’s help page. Ads will not be based directly on your Gmail inbox or Photos library, but summaries, excerpts, and inferences drawn from connected apps may be used to show you ads if you opted in.
Can advertisers see my Gmail or Google Photos?
No. Google states advertisers will not be able to access users’ connected apps, and it has announced no new advertiser controls or targeting options for this test.
How do I stop AI Mode ads from using my connected apps?
Disconnect the apps. Connecting content apps to Search Services is opt-in, and Google says you can adjust your settings, including disconnecting content apps, at any time.
When will AI Mode ads with Personal Intelligence roll out widely?
Google has not said. As of the October 16, 2026 confirmation, this is a limited test with undisclosed size and regions, and no announced wider rollout.
