Google quietly uses AI to rewrite news headlines! The media industry responds: this is equivalent to deciding what journalists should say.

Google is testing a highly controversial feature—using AI to directly rewrite news headlines in search results, replacing the carefully crafted titles created by media outlets. This is not a minor truncation or simplification, but a complete regeneration of a headline. The most disturbing case: a product review with a critical tone in the original title was rewritten by AI to appear as if it was recommending the product.

AI rewritten headline: “The Huge Impact Behind the ‘Small-Scale Experiment’”

According to a report by The Verge’s senior editor Sean Hollister, Google has begun replacing original headlines written by publishers with AI-generated versions in traditional search results (commonly known as “10 blue links”). Previously, Google had conducted similar tests in Google Discover and turned it into a permanent feature in January 2026.

Three Google spokespersons confirmed the existence of this test to The Verge, but described it as a “small-scale” and “limited” experiment that has not yet been approved for a full rollout. Google stated that the goal is to “better match headlines with users’ search intent and facilitate interaction with web content.”

Critical piece becomes recommendation: how a headline rewrite distorts the original news intent

The most widely cited case comes from an article by The Verge itself. The original title was: “I used the ‘cheat on everything’ AI tool and it didn’t help me cheat on anything”—which is clearly a critical review.

But Google’s AI rewrote the title to just five words: “’Cheat on everything’ AI tool.” For search users who haven’t read the original article, this looks completely like a product recommendation. Other cases include warning titles about the inability to replace LEGO Smart Bricks batteries and experience reports about Disney park robots, all rewritten into vaguer or more positive versions.

From “experiment” to “permanent feature”: A precedent set by Google Discover

What alarms media operators the most is the precedent set by Google Discover. At the end of 2025, Google began testing AI-rewritten headlines in Discover, also referred to as an “experiment.” But just about a month later, Google announced it as a permanent feature, claiming it “performed well in terms of user satisfaction.”

Hollister from The Verge bluntly warned: “You shouldn’t assume Google won’t roll out this feature on a larger scale, because Google initially told us that the AI headlines in Discover were just an experiment.” Notably, AI-generated headlines in Google Discover do indeed carry a disclaimer stating “generated by AI, may contain errors”—but this text is hidden under a “see more” button.

A dual crisis for media and SEO industries

For news publishers, the impact of this feature is a multi-layered compound blow. Referral traffic from Google Search has already significantly declined due to AI Overview—according to industry data from March 2026, AI Overview now appears in 30% to 45% of informational searches, with publishers experiencing a drop in organic traffic of 30% to 60%. The traffic brought by AI Overview’s cited links accounts for less than 1% of total traffic.

Headlines are the last tool publishers have to attract clicks in search results. If even this final control is replaced by AI, publishers’ presence in the entire search ecosystem will be further diluted. For SEO practitioners, industry advice has begun to shift: avoid using easily disassembled segmented structures in headlines, place the most important keywords at the very front of the title, and include the year to increase the chances of retaining the original headline.

Google’s contradictory stance: “No generative AI” but unclear alternatives

A Google spokesperson made a seemingly reassuring but actually vague promise: even if formally launched, “generative models will not be used, nor will generative AI be employed to create headlines.” However, Google did not explain how to generate entirely new headline text without using generative models—while the rewritten cases appearing in current tests clearly exceed simple truncation or rearranging of the original text.

A deeper issue: Who has the authority to frame the news?

At the heart of this controversy is a fundamental question about information power. When journalists write headlines, they are making editorial decisions about framing, emphasis, and accuracy. An AI system that rewrites these decisions without consent effectively becomes an invisible editorial barrier between journalists and readers, and is completely unaccountable.

Ben Schoon, an editor at 9to5Google, stated: “If Google can decide to display AI-generated headlines, it may completely distort the content that publishers actually released. This destroys an important element of the web for publishers and site owners.”

For the entire open web ecosystem, Google’s role is shifting from “neutral librarian” to “active editor of world information”—and this shift is occurring quietly without any public discussion or authorization.

This article “Google Search Quietly Uses AI to Rewrite News Headlines! Media Industry Pushback: This is Equivalent to Telling Journalists What to Say” first appeared in Chain News ABMedia.