Human-factor accidents are a major barrier to improving intrinsic mine safety and advancing intelligent transformation. They are characterized by high frequency, concealment, cascading effects, and preventability through intervention. As mining operations evolve toward efficient collaboration and intelligent operation, higher requirements are placed on dynamic individual risk identification, feedforward prediction, and closed-loop intervention. Based on the integration of embodied intelligence and system intelligence, a scientific concept of intelligent prevention for dynamic individual risk of human-factor accidents was proposed. The concept was designed to establish an intelligent prevention system with continuous learning, adaptation, and governance capabilities. Five major challenges were identified in multi-agent collaborative environments, including fragmented information acquisition, limited accumulation of data engineering resources, poor interpretability of modeling mechanisms, insufficient auditability of strategy execution, and inadequate governance support for system collaboration. A task-driven scenario benchmark, a distributed multi-agent intervention mechanism, and a continuously evolving paradigm for multi-source data alignment were constructed. An integrated technical framework was then established along the main pathway of “environmental perception–behavior prediction–system collaboration–strategy intervention.” Five key scientific issues were further identified, including risk evolution modeling, trustworthy reasoning outputs, near-miss data engineering, individual profile updating, and closed-loop strategy execution. Accordingly, five research directions are proposed, including dynamic operating-condition mechanism modeling, multi-scale perception–decision collaboration, robust model validation and generalization, transferable deployment of platform capabilities, and integrated governance of institutional mechanisms and intervention strategies. These directions accelerate breakthroughs in core technologies and systematic deployment. They provide theoretical support and an engineering pathway for improving mine operational safety, advancing the intelligent transformation of high-risk industries, and establishing an intrinsically safe national industrial system.