Meta has created its own AI-generated clickbait news feed

Facebook has long been full of clickbait article feeds. Now, Meta creates its own clickbait articles using AI.

The standalone Meta AI app now has a “For You” section that populates a list of clickbait-style stories for you to read. But the themes, images and text are all created by AI – and are as questionable as you would expect from works created with AI.

Apply Meta AI first Fired In April 2025 with a focus on the public Discover feed Show AI-generated photos and conversations from other users (who often seemed unaware they had been made public). That’s all gone. The app now has a standard chatbot interface, as well as a For You page that has been around for at least a few months, displaying a collection of suggested articles that, when clicked, generate entire “stories.”

When targeted at me, a London-based reporter, the prompts were strongly British, and included topics such as tea, manners, pubs, the royal family, football – sorry, football – and, of course, the art of waiting. Suggested stories included ‘The royal butler finally settles the milk-first debate’ (apparently tea goes first), ‘The psychology of joining a queue without knowing why’, ‘Anatomy of a devastating British raspberry’ and ‘Inside the extreme sport of visiting every pub in the UK’. Some were even less logical, such as “When a little pickle means complete disaster.”

Meanwhile, my colleague appears to have been firmly placed in the luxury watch lover category by the algorithm. His feed suggested stories called “My Bogus Rolex Experience” and “The Brutal Calculations Behind the Rolex Queue Illusion.”

The AI-generated text reads like bloated filler, offering little substance beyond repeatedly restating the premise of the claim. Sources were also non-existent.

I tried to trace where these “stories” originated. The story of the royal butler tea appears to go back to a 2018 BBC Three comedy series called Miss NetherlandsThe film follows a fictional beauty queen from a small Dutch town as she travels to Britain and learns “how to be posh and elegant” from the real former royal butler Grant Harold. Meanwhile, the “Rolex Experience” story seemed like a complete fabrication, created in our chatbox as a first-person story without a byline, after a bit of the usual buzz that occurs when creating a chatbot. Other stories relied on vague references to unnamed experts or fictional research.

When I tapped the same cards more than once, the stories generated stayed within the approximate boundaries of the prompt and were all clear versions of the same thing, just slightly different. Typing the same title into a separate conversation elicited a completely different response. The clearest gift came from my chat history. It showed the hidden and suggested prompts that were supposed to lead to the creation of the articles. Someone started:

“You are a helpful conversational assistant. The user responds to a proactive summary card that was shown to them. The context of the card below provides background on what prompted the user to send the message,” followed by what appear to be references to internal instructions, information and metadata.

The articles had pictures attached. A lot of it was harmless – a nice mix of cartoony people, scenery, and food. But some depicted real people, including public figures, and were full of errors. “Who will really pay for the royal family in 2026?” Two of Queen Elizabeth II appear, although she died several years ago and only exists as one person.

Around the cloned queen were people who appeared to be approximations of other royals: Princess Kate’s face vaguely to the left, a strange attempt at Prince William at the back, and a sort of King Charles in the middle bearing an exaggerated resemblance to his late father. Other images had ordinary AI-like hands and impossible bodies tilted at unnatural angles. In fact, one of the photos turned out to be a GIF of an older couple dancing and doing arm movements that a human body couldn’t do.

It wasn’t clear whether the app should be able to generate AI-powered images of real people based on meta-images, or if they’re just a bit opaque. rulesBut it was so. Previously for the company He said She wants “people to know when they see AI-generated posts” and to do so Automatically adds labels to some user-generated content When artificial intelligence is discovered. However, there was no clear indication or label in the feed or articles that any material was generated by artificial intelligence.

Meta declined to answer many of my questions about the purpose of the feature, whether the company is looking at news or fiction, what safeguards are in place, and whether images of real people and public figures comply with its AI content policies.

“We are testing a daily feed that proactively shares tips, content and recommendations tailored to your interests,” Tracy Clayton, a Meta spokeswoman, said in a brief statement. “The goal is to suggest what’s most important to you — like fitness tips, meal plans, or other ideas — before you even have to ask.”

Clayton later sent an almost identical “updated” statement, in which he mysteriously removed the word “proactive.”

A third statement from Clayton came later in the day: “This was a test for a limited number of users and will be deprecated. Meta has no plans to move forward with this feature.”

This leaves me with additional questions. How limited was this test if at least three of my classmates other than me were in Edge Did you have access to the same feature that serves AI clickbait? What does “proactive” even mean? And of course, who asked for any of this in the first place?