Here’s a risk worth naming before you scroll your next feed: an AI can now invent a “news” story about you, attach a fake photo, and serve it without ever telling you it’s fabricated. That’s exactly what Meta built. According to The Verge AI, the standalone Meta AI app rolled out a “For You” section that generated full clickbait-style articles where the topics, images, and text were all AI-made, with no sourcing and no labels. After The Verge AI started asking questions, Meta said it would kill the feature. But what it revealed about where AI content is heading is the part you should pay attention to.
What Meta actually shipped
The Meta AI app launched in April 2025 with a public “Discover” feed of AI images and user conversations. That’s gone. In its place sits a chatbot plus a “For You” page that suggests article prompts. Tap one, and the app spins up a complete “story” on the spot.
The Verge AI’s Robert Hart, reporting from London, got prompts that were aggressively British: tea, pubs, royals, and the art of queuing. Sample headlines included “A royal butler finally settled the milk first debate” and “The psychology of joining a queue without knowing why.” A colleague got slotted into a luxury-watch bracket with pieces like “My fake Rolex experiment.”
The text read like filler that restated the prompt over and over. Sourcing was nonexistent. When Hart tried to trace one story, it appeared to be a complete fabrication, generated live in the chat box with no byline.
Why this matters
The danger isn’t bad writing. It’s the missing guardrails.
- No labels. Meta has publicly said it wants “people to know when they see posts that have been made with AI.” The Verge AI found no clear AI label anywhere in the feed or the articles.
- Real people, fake images. One image for “Who really pays for the royal family in 2026?” showed two Queen Elizabeth IIs, despite her death years earlier. Others rendered rough approximations of Kate, William, and King Charles, complete with the usual AI tells like impossible hands.
- No sourcing. Stories leaned on unnamed experts or invented research, with one apparently tracing back to a 2018 BBC comedy series rather than any real reporting.
The giveaway came from chat history, which exposed the hidden system prompt feeding the generator: “You are a helpful conversational assistant. The user is responding to a proactive feed card…” In other words, the “news” was a chatbot performance dressed as journalism.
The walkback
Meta’s messaging shifted fast. Spokesperson Tracy Clayton first called it a test of “a daily feed that proactively shares tips, content, and recommendations.” A second statement quietly dropped the word “proactively.” A third landed later: “This was a test for a limited number of users and it will be deprecated. Meta has no plans to move forward with this feature.”
The Verge AI pushed back on the “limited” framing, noting that at least four of its own staff had access. Meta declined to answer most of the harder questions, including whether it considers the output news or fiction and whether AI images of real public figures comply with its own policies.
What stands out here
What’s striking is that a company the size of Meta shipped synthetic news with no label, no sourcing, and AI images of real royals, then only retreated once a reporter started digging. The feature is dead, but the appetite to fill feeds with auto-generated “content tailored to your interests” clearly isn’t.
If you build, moderate, or publish online, treat this as a preview. A few defensive moves worth making now:
- Assume feed content can be synthetic by default, even inside apps from major platforms, and check for a real byline or source before trusting it.
- Don’t rely on platforms to label AI content. This case shows the labels can simply be absent.
- Watch the images, not just the text. Fabricated photos of real people are already slipping past stated policies.
Meta walked this one back. The bigger question The Verge AI leaves open is who asked for AI-generated news in the first place, and how many similar experiments are running that no reporter has caught yet. You can read the full account at The Verge AI.