Asger Risom
Did I even write this headline?
A quiet experiment inside Google Discover is stirring unease across newsrooms. Android users have begun seeing altered headlines that don’t match the ones journalists wrote, raising concerns about confusion, click-through rates and who controls the first impression of a story.
Reports from multiple sources describe the same trend: auto-generated headers that shrink, distort or completely misrepresent the meaning of published articles.
Misleading mini-titles
Several of the AI-produced headlines highlighted by The Verge include provocative or inaccurate phrasing. One read “BG3 players exploit children,” even though PC Gamer’s actual reporting focused on a glitch in Baldur’s Gate 3 that lets players clone in-game child characters. Another Discover headline, “Steam Machine price revealed,” contradicted Ars Technica’s piece, which explicitly stated that Valve had not announced pricing.
Shortened versions have also stripped away key context. The Verge noted how “Microsoft developers using AI” reduced a straightforward story to a hollow statement. In other cases, fragments such as “Schedule 1 farming backup” or “AI tag debate heats” surfaced with no intelligible context.
According to The Verge, some headlines appeared as little more than four-word clickbait, while still displaying the original publisher’s name beside them.
Questions of control
Inputmag reported that Discover now shows a small label saying certain elements are “generated with AI, which can make mistakes.” But the notice only appears after tapping deeper into the interface, leaving many users unaware that Google—not the newsroom—authored the headline.
Editors quoted by The Verge argued that the practice undermines their ability to frame their own stories. They warned that readers could assume these distorted titles were intentionally crafted by the publication, eroding trust already strained by AI-driven changes to search and discovery.
The Danish report raised similar concerns, noting that a misleading header can push readers toward the wrong conclusion before they ever open the article.
Google’s explanation
Google told The Verge the feature is “a small UI experiment for a subset of Discover users” aimed at making topic details easier to scan. But the company did not clarify why rewritten titles frequently diverge from the meaning of the underlying reporting.
The experiment lands amid wider industry criticism that Google’s AI products—from search overviews to image outputs—are reshaping how information surfaces, often at the expense of publishers. As both outlets noted, the test also prompts broader questions about accuracy, transparency and the future of editorial control in algorithmic feeds.
Sources: The Verge, iNPUT
