Meta is adjusting the way its artificial intelligence suggests prompts after a series of invasive personal questions sparked a privacy outcry. The controversy began when Instagram user Kalie Robins shared a video showing how Meta AI reacted to a clip of her and her daughter singing in a car. After posting the video across platforms, Robins was met with an automated suggestion asking who the child passenger was. Upon clicking the prompt, the AI began synthesizing data from various sources to identify her children’s ages and pinpoint where the family resides.
The experience became even more unsettling for Robins when the chatbot surfaced photographs of her daughters that had been posted previously, including one image she believed had been deleted from the platform years ago. This revelation highlighted how deeply the AI integrates information from not only a user’s own history but also posts made by their relatives, creating a comprehensive profile that felt far too intimate for many observers.
In response to the incident, Meta spokesperson Dina El-Kassaby admitted that the company missed the mark and stated that the system should never have prompted a user with such specific personal queries. While Meta maintains that its AI only utilizes information that is already accessible to those viewing the content, they acknowledged that these particular suggestions were inappropriate and claimed that a fix has since been implemented to prevent similar occurrences involving sensitive personal topics.
This latest stumble comes as Meta continues to integrate its AI assistant throughout Facebook, Instagram, WhatsApp, and Messenger. The company has faced several hurdles while rolling out these tools, including a decision in July to scrap a feature allowing users to create deepfakes of others following intense public backlash. As Meta pushes further into generative AI, it faces increasing pressure to balance helpful automation with strict boundaries regarding user privacy and digital safety.
