Meta announced changes to its suggested AI prompts following public backlash over a viral video showing its chatbot asking invasive personal questions about a user’s children. Instagram user Kalie Robins demonstrated that after cross-posting a video, Meta AI suggested prompts asking to identify her child, details about her family’s location, and displayed previously deleted photos gathered from family members’ accounts.
In a statement to The Verge, Meta spokesperson Dina El-Kassaby admitted the company “missed the mark,” stating the feature should not have prompted users with such personal queries. Meta explained that the feature was designed to offer context on posts based on data the user already had access to, but confirmed it has now disabled prompt suggestions related to personal topics.
This incident adds to ongoing privacy concerns regarding Meta’s platform-wide integration of AI across Facebook, Instagram, WhatsApp, and Messenger. It follows a similar retreat in July when Meta pulled an Instagram feature that permitted users to generate AI deepfakes of other individuals.
Why it matters
Underscores severe privacy risks when AI models automatically parse and surface cross-platform personal and historical user data.
Highlights growing consumer pushback against proactive social media AI prompts, forcing platforms to limit personal inference features.
Signals potential regulatory and public relations exposure for consumer tech firms deploying unprompted AI interactions.
Source: theverge.com
