Meta made its own AI-generated clickbait news feed - The Verge

In June 2026, it was discovered that Meta’s standalone AI app was generating a feed of clickbait-style stories populated by AI-generated text and images. These stories, which targeted users with specific interests, frequently depicted real public figures and the British Royal Family with significant factual errors and visual inaccuracies.

What happened

The Meta AI app featured a "For You" section that proactively suggested prompts for users, which, when tapped, generated entire articles. These pieces lacked substance, sourcing, and bylines, often relying on fabricated narratives or vague references to unnamed experts. One generated article titled "Who really pays for the royal family in 2026?" included an AI-generated image depicting two versions of the late Queen Elizabeth II, alongside approximations of other royals that displayed common AI artifacts like impossible anatomy and unnatural movement.

Despite Meta’s stated policies regarding the labeling of AI-generated content, the articles in the app feed lacked any clear indication that the material was synthetically created. The text itself appeared to be "puffy filler" that restated the premise of the prompt without providing genuine information. When prompted, the AI could generate different versions of the same story, reinforcing the lack of factual grounding. Following inquiries regarding the nature of this feature, the source of the generated content, and compliance with internal AI policies, a Meta spokesperson stated the feature was a limited test and would be deprecated, with no plans to move forward. The company declined to clarify its internal safeguards or the specific purpose of the AI-generated feed.