Updated
Updated · The Verge · Sep 11
Meta Changes AI Prompts After Chatbot Probed User's 2 Daughters
Updated
Updated · The Verge · Sep 11

Meta Changes AI Prompts After Chatbot Probed User's 2 Daughters

3 articles · Updated · The Verge · Sep 11

Summary

  • Meta said it has fixed the issue behind AI prompt suggestions on Facebook after a viral video showed the chatbot asking invasive questions about a user’s two young daughters.
  • One prompt under Kalie Robins’ video asked, “Who is the child passenger?” and, after she clicked it, Meta AI surfaced details from her and relatives’ posts, including follow-up questions about her children’s ages and where the family lives.
  • Meta spokesperson Dina El-Kassaby said the company “missed the mark” and that the feature should not have suggested such personal questions, adding the responses were generated from information the user could already access on the platform.
  • The episode adds to scrutiny of Meta’s AI rollout across Facebook, Instagram, WhatsApp and Messenger, coming two months after it pulled an Instagram tool that let users create AI deepfakes of others.

Insights

If Meta's AI can retrieve your deleted family photos to answer questions, what else is hiding in your digital footprint?
How did a simple social media chatbot manage to piece together a child's exact identity and location from supposedly deleted data?