The control of robotic lower limb prostheses is shifting from a “one-size-fits-all” paradigm to highly individualized approaches. This is because artificial intelligence, and particularly machine learning, now offer powerful tools to adapt how a prosthesis behaves to the unique needs of its user. Control algorithms capable of capturing each user’s specific locomotion patterns are making the initial post-operative rehabilitation phase after having a prosthesis fitted easier, ultimately improving the quality of life and comfort of individuals with lower limb amputations. Additionally, real-time intention detection and human-in-the-loop optimization are showing a great deal of promise as data-driven methods for aligning a prosthesis’s behavior with the user’s intentions. However, significant challenges remain. Accuracy, real-time responsiveness, efficient tuning, and adaptability to the learning curve of users are all issues that still need to be addressed. Legal compliance is posing additional hurdles, as coordination between the Medical Devices Regulation and the recently passed AI Act remains unclear. This regulatory ambiguity could hinder innovation and market access for robotic prostheses. Hence, this chapter presents advanced control strategies for robotic lower limb prostheses, with a particular focus on machine learning approaches. We discuss the opportunities and challenges associated with developing the next generation of assistive control algorithms and present potential pathways toward addressing regulatory requirements given Europe’s new legislation. Our goal is to promote both innovation and compliance in today’s rapidly evolving field of prosthetic control systems.