Meta Adjusts AI Chatbot After Invasive Questions
Meta is updating its AI chatbot features following reports that the system posed inappropriate questions about a user's minor children.
Meta has announced changes to the prompt suggestions within its AI chatbot after a viral incident highlighted how the system asked intrusive questions about a woman's young daughters. The company acknowledged that the feature "missed the mark" and stated that it never should have prompted users with such personal inquiries.
The issue came to light after an Instagram user shared a video showing the AI suggesting questions about a child passenger in a clip she posted. Meta confirmed it has fixed the specific issue causing these prompts and clarified that the system only accesses content the user is already permitted to view.
Viral video exposes privacy gap
The controversy began when Kalie Robins cross-posted a video of herself and her child singing in a car to Facebook. Beneath the clip, the AI suggested the prompt, "Who is the child passenger?" When Robins selected the suggestion, the system appeared to aggregate information from her previous posts and those of her relatives. It then generated further questions regarding her children's ages and her family's location, while also displaying photos of her daughters, including one she claimed had been deleted years earlier.
Company explains data access limits
In response, Meta spokesperson Dina El-Kassaby told The Verge that the AI system does not retrieve private data outside of what the user can already see. "Meta AI returns responses containing content that the querying user can already access," El-Kassaby explained. She noted that if a user cannot view a specific post, the AI will not include it in its response. In this instance, the company believes the AI pulled information from videos where Robins had previously named her children.
Pattern of AI feature reversals
This incident follows a previous controversy in July, when Meta removed a feature allowing users to create AI deepfakes of other Instagram profiles after facing significant public backlash. The AI assistant is currently integrated into Facebook, Instagram, WhatsApp, and Messenger, where it offers services like answering questions and generating images. While the company states the feature was designed to help users find information on topics they find interesting, the recent changes reflect a tightening of safety guidelines around personal data.
