Meta's AI app generated a clickbait news feed with fake stories and errors
Meta's AI app created a For You feed of AI-written clickbait, then pulled it after inquiries.
Meta's standalone AI app introduced a 'For You' section that populates a feed of AI-generated clickbait articles. The stories, triggered by prompt cards, covered topics like tea etiquette, queue psychology, and a fake Rolex experiment. The generated text lacked substance and sourcing, often repeating the premise without adding value. Images attached to articles were riddled with AI artifacts, including a depiction of two Queen Elizabeth IIs despite her death years earlier, and other royals with distorted features. Internal chat history showed hidden instructions for the AI, including a system prompt describing the assistant's role and metadata references.
Meta confirmed it would pull the feature after The Verge raised questions, but the incident highlights ongoing challenges with AI-generated content moderation. The feature appeared to aim for engagement over accuracy, producing fabricated narratives and dubious assertions. For tech professionals, this underscores the risks of deploying generative AI in consumer-facing content without robust safeguards. It also raises concerns about how platforms handle AI-generated misinformation and the need for transparency in AI-driven editorial decisions.
- Meta's AI app had a 'For You' feed that generated clickbait articles with no sourcing or verified facts.
- One image showed two Queen Elizabeth IIs, and a 'Rolex experiment' story was entirely fabricated.
- Internal chat logs revealed hidden system prompts and metadata used to generate the content.
Why It Matters
Meta's botched AI feed shows how easily generative AI can produce low-quality, misleading content at scale.