Google rewrites headlines in search results with AI

When a headline appears in Google search results, it may no longer belong to the website behind it. Google is currently testing AI-generated changes to search result headlines, as the company has now confirmed. The known cases involve journalistic online articles.

Authors at the US tech magazine The Verge noticed this in recent months. In two cases, their articles appeared in Google Search with headlines that were not their own. While it is not uncommon for a different headline to be chosen than the actual one of the article – these alternative headlines are provided by the authors or media houses themselves so that an article can be found better via Google.

Google AI changes the meaning of headlines

Most importantly, the authors' headlines also correspond to what is actually in the text, which runs under the author's name. This was not always the case with the headlines reformulated by Google. For example, the headline "I used the ‘cheat on everything’ AI tool and it didn’t help me cheat on anything" was changed to "Cheat on everything’ AI tool". Instead of a clear criticism of the AI tool Cluely, the headline suddenly sounded more like a presentation or marketing of the tool. In a random Google search from Germany, this and another affected article are returned with their original headlines (as of March 22, 2026, 12:33 PM). However, it is quite possible that the changed headlines are only displayed in certain regions or to certain users.

Google admitted to The Verge that the two headlines were changed by AI, as were the headlines of other websites. This was part of a "small" experiment. According to Google, AI was used to search for content on the respective websites that could correspond to a relevant search query. The aim was to better adapt the headlines to the search queries in order to "facilitate engagement with web content."

Google emphasizes: "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." The search engine giant does not say how the solution would look instead.

AI interventions in Discover are already standard

The Verge now fears that Google's "experiment" could quickly become the standard – as it has in the US with Google Discover. As part of another experiment, Google replaced the titles of news articles in the Discover collection in the US at the end of 2025 using AI technology – and sometimes produced hair-raising errors. Discover is also available in this country; Google collects articles from various online media here and thus offers an overview of the latest news as well as articles worth reading, regardless of their source.

Users can scroll through these in the Google app or on the mobile site; clicking on one takes them directly to the respective news pages. However, recently the Discover texts for US users often completely misrepresented what was in the linked text of the respective news page. Without the affected media having any influence on this. This practice has now become standard in the US, as Google told The Verge in January, because it contributes well to user satisfaction. Whether and how user satisfaction was measured remains unclear. Incorrect headlines or Discover texts likely generate many clicks, but also many frustrated readers. For Google, only the former likely counts – because an incorrect Discover text is usually attributed to the medium of the article behind it.

For journalists and internet authors, these developments are a major problem. Because what is replaced by Google's AI was previously carefully formulated with a lot of work – with the aim of arousing interest in the article, but also of not creating false expectations. If Google now undermines this with faulty Discover texts or headlines in Google Search, the reputation of the media houses can suffer.

(nen)