Headlines Generated by Google’s AI Are Here to Stay.
Last month, Google informed online publishers that it was beginning to test AI-generated headlines in Google Discover. Hand-crafted article headlines were replaced with shortened versions developed by Gemini. Some journalists, as expected, were unhappy, but the company now states that AI-generated headlines are no longer an experiment—they’re a “feature.”
When testing first began, the results ranged from poorly worded to outright misinformation. For example, one AI-generated headline promised “Steam Machine Price Revealed,” though the original article made no such claim. Another read, “BG3 Players Exploit Children,” which sounds serious until you click the link and see that it’s about a clever way to recruit invincible party members in Baldur’s Gate 3 (which, to be fair, at one point involves turning child NPCs into sheep).
At the time, Google stated that the test was a “small user experience experiment for a subset of Discover users” and simply changed the way it displays AI-generated previews. These previews, introduced last October and featuring short AI-generated article summaries, sometimes including an AI-generated headline, were previously hidden beneath the original headline. However, during the test, they were moved to the top, and the original headline was removed entirely.
For a while, it seemed Google was ready to abandon AI-related headlines , but now the company says it intends to redouble its efforts. In a statement to The Verge , Google announced that its AI-related headlines are no longer in testing and are now a fully-fledged feature. The company didn’t elaborate on the reasons, but noted that the update “is working well in terms of user satisfaction.”
When 9to5Google reached out to the publication for more details, they responded: “The overview headline reflects information from multiple sites and is not a rewrite of a single article’s headline.” Well, in my case, that wasn’t quite true: when I first wrote about this “experiment,” I hadn’t yet encountered any of the AI-generated headlines. But today, while scrolling through my Google Discover feed (to see yours, swipe right from the home screen on an Android phone or scroll down in the Google app), I finally saw them for myself. To Google’s credit, these AI-generated previews do appear to aggregate multiple sources as advertised—you can see them above the article link. However, they still highlight one specific article by linking to it and using its photo in the headline. This could easily lead users to believe the AI-generated headline was written by that specific publication.
This could have consequences for the publication or the author if the AI makes a mistake, which, as the disclaimer at the bottom of these AI previews acknowledges, is entirely possible. For example, The Verge reported seeing an AI Discover headline in an article on PCMag , a Lifehacker affiliate, that read, “The US has lifted its ban on the use of drones abroad,” even though the linked article specifically emphasized that headlines claiming this are “misleading.”
The AI-generated headlines I’ve seen personally aren’t all that bad, but as someone with over a decade of journalism experience, I question their usefulness. For example, “Starfleet Academy Is Full of ‘Star Trek’ References” is far less informative than the original headline, “One of ‘Star Trek: The Next Generation’s’ Strangest Species Gets a Second Life in Modern ‘Star Trek.'” Apparently, the phrase ” Star Trek has references to ‘Star Trek ‘” is more engaging or useful to the reader than simply pointing out a specific Star Trek reference?
Another example: “Anbernic Introduces RG G01 Controller.” Hopefully you understand what those letters and numbers mean, because this AI-generated headline completely obscures the context of the original headline: “Anbernic’s New Controller Has a Screen and a Built-in Heart Rate Sensor for Some Reason.”
I suppose I’ll have to get used to this future. The fact that I’m starting to see these headlines myself, despite not participating in the initial experiment, suggests they’re likely here to stay and will become available to more users. If you see something suspicious in Google Discover, this feature is likely already available to you, too.
How to check if your Google Discover headline was written by AI.
To check whether the suspicious headline was written by a human or not, try clicking the “Read More” button at the bottom of the article description and look for the “AI Generated” warning.
On the other hand, only about half of the articles in my Google Discover currently use AI-generated headlines, so this doesn’t affect all “content.” But this move still comes at a challenging time for journalists: according to Reuters , organic Google search traffic to test sites in the US fell by 38% between November 2024 and November 2025. And while Google Discover isn’t search, editors write headlines that way for a reason. Using a robot to overturn these decisions probably isn’t the best way to combat eroding trust in the media .
I’ve reached out to Google for comment on the AI news and will report back as soon as I hear back.