AI experiment rewrites article titles
Google Search is testing a new feature that uses artificial intelligence to alter headlines and website titles shown in search results. The move, first highlighted by The Verge, has reportedly been appearing over the past few months, with some news organizations noticing that headlines displayed in Search did not match the originals written by their editorial teams.
According to Google, the experiment affects a small number of results and is designed to identify content on a page that may serve as a more relevant title for a user’s query. The company says the goal is to better match titles to search intent and improve engagement with web content.
Google says no generative AI would be used
Google clarified that if the experiment were to expand, it would not rely on generative AI models to create entirely new headlines. Instead, it would select alternative text already present on the page. The company emphasized that it would not be “creating headlines with gen AI” in a full rollout.
Still, examples shared publicly show noticeable changes. One article originally titled “I used the ‘cheat on everything’ AI tool and it didn’t help me cheat on anything” was shortened in search results to “Cheat on everything AI tool.” Another headline, “Microsoft is rebranding Copilot in the most Microsoft way possible,” appeared instead as “Copilot Changes: Marketing Teams at it Again.”
Concerns over accuracy and publisher control
It is common practice for publishers to use separate SEO titles that differ slightly from visible headlines. However, those versions are typically created by authors or editors. While Google has long truncated headlines to fit display limits, fully rewriting or substituting them represents a significant shift.
Critics argue the experiment risks misrepresenting content by altering tone or intent. Headlines are often carefully crafted to convey nuance, context, or editorial voice. An AI-driven rewrite could oversimplify or distort that message.
Traffic pressures and broader AI rollout
The test also comes at a time when referral traffic from Google Search has reportedly declined for many publishers. AI-generated summaries and “source” links in search results have not compensated for those losses, with some reports suggesting such links account for less than 1% of overall traffic.
Google previously tested AI-driven headline adjustments in Google Discover, where the company said the changes performed well in terms of user satisfaction. Applying similar logic to Search, however, raises broader questions about how much editorial control publishers retain over how their work is presented.
While Google frames the experiment as a way to better align results with user queries, critics contend it alters a foundational element of the web: the publisher’s authority over their own titles. Whether the feature expands beyond limited testing remains to be seen.
