The development of artificial intelligence poses significant ethical, social, and environmental challenges, particularly regarding the reproduction of structural biases —sexist, racist, ableist— and the invisibility of the human labour that underpins AI systems. This debate will examine the need for effective oversight, more transparent and auditable models, and accountability frameworks for those implementing AI. From multiple perspectives —from data architecture to digital justice in global contexts— the session will also explore the ecological costs of large-scale model training and the urgent need to imagine sustainable, decentralised, and culturally diverse technological infrastructures.