Meta is modifying its AI chatbot prompts following reports that the assistant generated invasive questions about a user's children. The issue surfaced after Instagram user Kalie Robins shared a video showing the AI suggesting prompts like 'Who is the child passenger?' beneath a clip of her and her daughter.
According to the report, the AI aggregated information from the user's previous posts and content shared by her relatives. The assistant also displayed photos of the children, including one that the user claims to have deleted years prior. Meta spokesperson Dina El-Kassaby stated that the company failed to meet its standards and that such prompts should not have occurred.
Meta has integrated its AI assistant across Facebook, Instagram, WhatsApp, and Messenger to assist with tasks like writing posts and generating images. The company claims to have fixed the technical issue that caused the AI to suggest personal topics. El-Kassaby noted that the feature is intended to provide information on topics of interest based on data already accessible to the user.
This incident follows a previous controversy in July, when Meta withdrew a feature that enabled the creation of AI deepfakes of other Instagram users. The company faced significant criticism at that time, leading to the removal of the tool. Meta maintains that the current AI assistant functions by pulling from information that is already available to the user, such as videos where names were mentioned.
