Meta Scraps AI News Feed Test After Verge Finds Unlabeled Stories and 3 Colleagues Access
Updated
Updated · The Verge · Jun 6
Meta Scraps AI News Feed Test After Verge Finds Unlabeled Stories and 3 Colleagues Access
1 articles · Updated · The Verge · Jun 6
Summary
Meta said it will deprecate a limited test in its standalone Meta AI app after The Verge questioned a “For You” feed that generated full clickbait-style stories from suggested prompts.
The feed surfaced at least for months and lacked clear AI labels, even though Meta has said users should know when content is AI-made; the company also would not clarify whether it viewed the output as news or fiction.
Stories were often thin or fabricated, with no sourcing and shifting text on repeat taps, while hidden prompt text in chat history suggested the articles were being generated on demand from internal instructions.
Images attached to the stories included obvious errors — such as two Queen Elizabeth IIs in a royal-family graphic — and appeared to depict real public figures despite questions over compliance with Meta’s own AI-content rules.
Meta first described the feature as a daily feed sharing tailored tips and recommendations, then said it had no plans to move forward, though The Verge reported at least 3 colleagues could access the supposedly limited test.