Meta made its own AI-generated clickbait news feed
Meta’s AI app generated clickbait-style story prompts and articles, then said it would deprecate the feature after questions from The Verge.
Intelligence analysis by GPT-5.4 Mini
The Verge found that the Meta AI app’s “For You” feed was serving AI-made article prompts and generated stories that often read like empty filler, with questionable sourcing and odd results. Meta later said the test was limited and would be deprecated.
Meta’s app was acting like a pretend newspaper machine, making up little story ideas and articles based on guesses about what people might like. Some of the stories were silly, and some had obvious mistakes, like drawing the wrong royals.
Analysis
What Meta built
The Verge reports that the standalone Meta AI app has a “For You” section that suggests clickbait-style story prompts and then generates full stories when tapped. The prompts were personalized in a strange way, including British-themed topics for the reporter in London and luxury-watch stories for a colleague.
What the output looked like
The generated text was described as thin and repetitive, with little beyond restating the prompt. The sourcing was weak or nonexistent. In one case, the story seemed to trace back to a BBC Three comedy series, while another appeared to be a fabricated first-person “experiment.” The feed also produced images with obvious AI mistakes, including duplicated public figures and other visual errors.
Why the article is concerned
The Verge says there was no clear labeling in the feed or the generated stories showing that the material was AI-generated, even though Meta has previously said it wants people to know when they are seeing AI-made posts. The article also says Meta did not answer many questions about the feature’s purpose, whether the output should be treated as news or fiction, or how its safeguards work.
Meta’s statements changed over the day. First, a spokesperson described the feature as a test for a daily feed that would share tips, content, and recommendations. Later, Meta said the test was for a limited number of users and would be deprecated, with no plans to move forward.
The story is less about one awkward feature than about the broader pattern: Meta experimented with AI-generated feed content that blurred the line between recommendation, fiction, and news, then pulled back only after scrutiny.
Key points
- The Meta AI app had a “For You” feed that generated clickbait-style stories from AI-made prompts.
- The generated stories were often repetitive, weakly sourced, and sometimes nonsensical.
- Some outputs included images of real people and obvious visual errors, including duplicated Queen Elizabeth IIs.
- The Verge says the feed lacked obvious labels identifying the content as AI-generated.
- Meta later said the feature was a limited test and would be deprecated.
If used carefully, a feed like this could surface useful tips, recommendations, and topic ideas before someone even asks. Meta’s decision to call it a limited test and deprecate it suggests the company may be willing to back away from a feature that is too confusing or poorly controlled.
The test showed how easily AI feed products can blur the line between helpful suggestions and fake content. The lack of clear labeling, weak sourcing, and errors involving real people could erode trust if similar features are shipped more broadly.
