Google Reverses Course, Testing Ads Based on Gmail and Photos AI Summaries

For nearly a decade, Google maintained a strict firewall between its advertising network and the private contents of users' inboxes, retreating from an early practice of scanning Gmail for ad keywords in 2017 after sustained public backlash. Now, the company is returning to personal data for ad targeting, but it is doing so through a distinctly different mechanism. Rather than scanning raw text for keywords, Google is leveraging the summaries and inferences generated by its artificial intelligence tools.[2]

The shift was formalized on October 16, 2026, when Google quietly updated its privacy policy. The revised document states that Google Search may now process information users allow it to access in order to "personalize content, including ads." This creates a specific exception to the company's long-standing rule against using Google Workspace data for advertising, applying exclusively to users who opt into a feature called Personal Intelligence.[4][5]

Personal Intelligence is the connective tissue that links a user's private Google Drive, Gmail, and Google Photos accounts to the company's public-facing AI Mode and Gemini chatbot. When enabled by users aged 18 or older, it allows the artificial intelligence to answer queries using the user's own files. Under the new policy, the insights generated during those interactions can now dictate which sponsored products appear alongside the answers.[1][3]

Reversing a prior commitment

The October update represents a direct reversal of the company's stated position from earlier in the year. "There are currently no ads for people who choose to connect their apps with AI Mode," a Google spokesperson told Search Engine Land in March 2026.[1]

That firewall has now been lowered. Google confirmed on October 16 that it is actively testing personalized advertisements inside AI Mode for users who have enabled Personal Intelligence. The company has not disclosed the geographic scope of the test or the percentage of opted-in users currently seeing the new ad formats, though AI Mode ads currently run in English across 12 countries.[1]

Google maintains that the new system does not feed raw emails or private photographs directly to advertisers. Instead, the ad targeting relies entirely on the secondary inferences and summaries that the artificial intelligence generates to answer a user's prompt. The distinction separates the underlying data from the marketing profile, though the practical outcome for the user is similar.[1][2]

If a user asks the AI to summarize an upcoming flight itinerary buried in their inbox, the system processes that request and generates a response. Under the new testing framework, the AI can simultaneously use that travel context to serve targeted advertisements for hotels, rental cars, or tours at the destination, seamlessly blending private context with commercial placements.[2][4]

Boundaries and user controls

To mitigate privacy concerns, Google has implemented several boundaries around the new targeting capability. The company stated that it will not personalize advertisements based on sensitive categories, which include health information, religious beliefs, and sexual orientation. Advertisers do not receive any direct access to the connected Workspace data, and the targeting process occurs entirely within Google's internal servers.[1][3]

Google Workspace currently serves more than 3 billion users globally, making the potential pool for this targeted advertising massive, even if only a fraction of users opt into the Personal Intelligence layer. Because the feature requires an explicit opt-in, users who have never connected their private apps to Google's AI tools are not affected by the policy change.[3][4]

For users who have already enabled the feature but wish to opt out of the new ad targeting, the connection can be severed without deleting the underlying accounts. Users must navigate to their Google Account settings, open the "Data & privacy" tab, and locate the "Personalization settings" menu. From there, they can disconnect Google Workspace and Google Photos from Search Services.[3]

The cost of conversational search

The policy update highlights the evolving, and increasingly expensive, monetization strategy for generative artificial intelligence. As search engines shift away from returning simple lists of blue links and toward providing conversational, synthesized answers, the computational cost of running those queries has skyrocketed. Google's ad revenue, which routinely exceeds $100 billion per quarter, relies on finding new surfaces for commercial placements.[1]

Traditional keyword advertising struggles to map neatly onto conversational interfaces, forcing technology companies to find new ways to extract value from AI interactions. By tapping into the rich, highly specific context of a user's personal files, Google can offer advertisers a level of targeting precision that standard web searches cannot match.[1][4]

Digital marketing analysts note that the test currently offers no new controls for the advertisers themselves. Brands cannot specifically target users who have enabled Personal Intelligence, nor can they opt out of having their ads displayed in AI Mode. The system simply uses the enhanced personal context to better match existing Search and Performance Max campaigns to the user's implied intent.[1]

The exact rollout timeline for a broader launch remains undisclosed, and Google has not indicated whether the test will expand beyond its current limited scope. For now, the boundary between private inboxes and public advertisements remains intact—but only for those who decline the artificial intelligence's assistance.[1][3]