Publishers running Google AdSense ad intent units are now hosting AI-written content they never assigned, edited, or approved. Google has quietly turned on Gemini-generated mini-articles inside the popup dialog that opens when a reader clicks an ad intent link, anchor, or chip, and there is no setting to switch it back off.
Google’s own support documentation confirms the mechanic: clicking an ad intent element opens a modal where AI-written text now runs alongside the display ads that used to occupy that space alone. The company frames this as a quality upgrade, telling publishers its internal testing found that pairing the generated text with the ads made the dialog more useful to visitors and lifted revenue. Search Engine Roundtable, which surfaced the documentation change, noted that Google has published no opt-out mechanism for the AI content.
The consequence reaches well past a layout tweak. Ad intent units sit inside a publisher’s own article content, not in a separate ad slot served by an outside network. That means Google is now writing and publishing original text on someone else’s domain, under that publisher’s name, with no editor in the loop. A site that spent years earning a specific voice and a track record for accuracy suddenly has machine-written mini-articles slotted into the same reading experience, credited to nobody on staff.
That is where this becomes an E-E-A-T problem, not just a product update. Google’s own experience-expertise-authoritativeness-trust framework assumes a site’s credibility is built sentence by sentence, byline by byline, through content its owner actually stands behind. Text the publisher never wrote or fact-checked can now live inside that same page shell. A reader has no visual cue to separate a Gemini output from staff-written material, and Google’s quality systems, when they later crawl the page, may struggle to make that distinction either.
Nothing in Google’s rollout notes offers third-party verification of the new content’s accuracy, nor any explanation of how the generated text is sourced or checked before it publishes. The only evidence Google has offered is its own internal read on user experience and earnings, not an outside audit of factual quality or the brand risk this creates for the sites carrying it.
Marketer Bruno Ramos Lara posted screenshots on X on June 30 showing the feature live: roughly 500-word AI-written posts tied to whatever ad intent a visitor triggered, woven directly between the display ads inside the same popup. Google’s guidance to site owners was equally brief, telling them the content now appears automatically and there is nothing for them to configure.
The missing toggle is the part publishers should sit with longest. Auto ads and other AdSense automation have typically shipped with dials publishers can turn, letting them block ad formats or placements that clash with their brand or their readers. Here, Google has folded AI-generated article content into the ad unit as a built-in default rather than a separate layer anyone can switch off. Any site already running ad intents is, as of this rollout, distributing Gemini-written material on its pages whether the publisher signed up for that or not.
Site owners running ad intent units should pull up their live pages this week and read exactly what Gemini is generating next to their ads, then treat any claims in that text as unverified until Google ships a way to review, correct, or disable it.
Search Engine Roundtable reported this AdSense change around June 30 and July 1, 2026, citing Google’s own AdSense documentation.
