According to AI at Meta, Meta has introduced a comprehensive multimodal dataset specifically designed for AI reading recognition tasks in real-world environments. The dataset combines video, eye gaze tracking, and head pose sensor outputs collected from wearable devices, facilitating the development of advanced AI models capable of understanding human reading behaviors in diverse settings. This resource is expected to accelerate research in human-computer interaction, personalized learning, and adaptive reading technologies by enabling more accurate reading activity detection and analytics (Source: AI at Meta, June 13, 2025).