Google has added a user-curation layer to its generative search results. On May 27, 2026, Duncan Osborn, Product Manager at Google Search, announced on the official Google blog that Preferred Sources, the feature that lets users build a personal list of trusted publishers, now feeds both AI Mode (Google’s conversational search experience) and AI Overviews.
The announcement introduces two related changes alongside Preferred Sources. A Highly Cited label will mark stories that other publishers frequently reference, rewarding primary reporting in a visible way. A new article carousel will surface multiple publisher perspectives on developing topics, giving readers a horizontal scan of coverage rather than a single synthesized answer.
Google reported that users have already selected more than 345,000 unique sources since Preferred Sources launched. That figure signals meaningful adoption and establishes Preferred Sources as a channel with real distribution weight rather than a cosmetic personalization option.
For publishers, this is a structural shift in how AI Overviews visibility is acquired. Until now, appearing in an AI Overview required satisfying Google’s algorithmic assessment of quality, authority, and relevance. That remains true. But Preferred Sources adds a second pathway: direct user selection. A reader who has added a publication to their list will see that publication’s content surface in AI Overviews when the query is relevant. The algorithm did not choose it. The user did.
The practical implication is that audience loyalty now converts into algorithmic presence in a way it previously did not. Publishers with strong direct relationships, newsletter subscribers who follow their brand, readers who actively seek them out, gain a distribution mechanism inside the AI layer of Google Search. This is not a small adjustment to how AI Overviews are assembled. It is the introduction of a user-pull signal alongside the existing algorithm-pull signal.
Google has also said it is working toward using Preferred Sources as a ranking signal across AI features in the future, which would extend the influence of user selection beyond personalized results.
The Highly Cited label carries a separate but related implication for SEO teams. If Google is surfacing stories that other publishers cite, derivative aggregators and summary-only pieces will be disadvantaged in the article carousel. The label functions as a proxy for primary reporting. Sites that break news, produce original data, or generate the coverage that others reference will earn the label; sites that republish or lightly reframe original work will not. In carousel real estate adjacent to AI Overviews, that distinction becomes a visibility gap.
SEO teams should treat this as two parallel action items. First, build or grow the direct audience channels, email, RSS, social follow, that encourage users to add the publication to their Preferred Sources list. Second, audit content operations to determine how much original versus derivative reporting the site produces. The Highly Cited label does not reward volume. It rewards being the record-of-source for a topic.
The announcement does not include an independent measurement of how much Preferred Sources currently shifts AI Overview citations in practice, nor does Google disclose what percentage of AI Overview sessions involve a user who has set Preferred Sources. Those numbers would clarify the size of the opportunity. What is clear is that Google has now formalized a user-driven path into the generative search layer, and publishers that invest in direct audience relationships will gain AI visibility through a channel that aggregators cannot replicate.
SEO and content teams should run a Preferred Sources audit now: query your own brand in AI Mode, check whether your source appears in AI Overviews when a reader would reasonably expect it, and set a benchmark before Google expands the ranking-signal application of Preferred Sources across more AI features.
Reported by the official Google Search Central blog on 2026-05-27, written by Duncan Osborn, Product Manager at Google Search.
