This thesis explores AI-assisted assembly systems for inclusive manufacturing through a comparative techno-legal analysis of data, cognitive support, and governance. Within the Industry 5.0 paradigm, manufacturing systems are increasingly expected not only to improve productivity and quality but also to promote human well-being, accommodate workforce diversity, and support more sustainable forms of human-machine collaboration. The research compares three AI-assisted assembly configurations selected according to two analytical variables: the type of data processed and the type of AI employed. The first system combines non-physiological task-related data with generative AI to provide accessible corrective instructions. The second relies on physiological and behavioural data with non-generative AI to estimate cognitive load and support reciprocal learning in human-robot collaboration.