Meta patches invasive AI suggestions

Meta is changing its AI-suggested prompts after one surfaced invasive questions about a user’s children; the assistant currently spans 4 apps—Facebook, Instagram, WhatsApp, and Messenger. In The Verge’s report on Meta’s response, the company says it “missed the mark” and fixed the feature after a prompt asked who a child in a video was, then led to questions about the children’s ages and home location; Meta says responses only use content the user can already access. Builders shipping assistants over personal archives should separate access permission from inference permission, a sharper boundary than Meta’s contributor-data pricing case.