Google has confirmed it is testing AI-generated headline rewrites in its traditional search results, a significant escalation of a practice that already became a permanent feature in Google Discover after barely a month of testing.
The company described the test to The Verge as “small and narrow,” language that closely mirrors what Google said in December 2025 when it began rewriting headlines in Discover. That earlier test was called “a small UI experiment for a subset of Discover users.” By January 2026, Google had reclassified it as a feature, saying it “performs well for user satisfaction.” The entire lifecycle from experiment to permanent change took roughly four weeks. The search test follows the same opening move.
Multiple media outlets observed rewritten headlines appearing in Google search results over recent months. In one case, an article originally titled “I used the ‘cheat on everything’ AI tool and it didn’t help me cheat on anything” was shortened in results to “‘Cheat on everything’ AI tool.” Another article was displayed with the headline “Copilot Changes: Marketing Teams at it Again,” phrasing that appeared nowhere in the original piece.
Later, Google told The Verge that the goal is to “identify content on a page that would be a useful and relevant title to a users’ query” and that the test aims at “better matching titles to users’ queries and facilitating engagement with web content.”
“If we were to actually launch something based on this experiment, it would not be using a generative model and we would not be creating headlines with gen AI,” Google added.
This is different from Google’s long-standing practice of rewriting title tags. Google has been adjusting how headlines appear in search for years using rule-based systems that pull from existing on-page elements like title tags, H1 headings, og:title meta tags, and anchor text. One study of over 80,000 title tags found Google changed 61% of them, while a follow-up study put the number at 76%. But those rewrites were always drawn from content that already existed on the page.
The new test uses generative AI to create entirely new text. In the Copilot example, the rewritten headline contained phrasing the article never used. That is a fundamentally different capability, and it raises different concerns.
However, none of the observed rewrites included any disclosure that Google had altered the original headline. Google’s title link documentation describes the inputs the system may use to generate titles but offers publishers no mechanism for opting out.
If AI headline rewrites become a feature in traditional search as well, publishers would lose control over how their content is presented across both of their primary Google traffic sources simultaneously.
The concern is not just about branding. Misrepresented headlines can distort the meaning of an article. A headline is often the only thing a user sees before deciding whether to click, and if Google’s AI changes the framing, the publisher bears the reputational consequences for words they never wrote.
