The increasing prevalence of AI-enabled systems demands robust control algorithms to manage complex interactions in dynamic, uncertain environments.  Cognitive science has formulated several principles underlying principles underlying human cooperation and decision-making. Despite each human pursuing individual goals, human groups excel at resolving mixed-motive challenges—situations involving both competition and collaboration—through mechanisms like shared intent, reward prediction, and adaptive behavior.  In this project, the computational principles of Active Inference will be combined with deep learning to enable cooperation between AI systems., to address key challenges like emergent behavior, conflict resolution, and decentralized control. The algorithms will be developed  for applications in robotics, human-AI systems, and smart infrastructures.