Artificial Intelligence has proven to be successful for the most diverse tasks and applications, attracting the interest of industries and investors. However, the application of common AI techniques to many real-world scenarios characterised by scarse resources is still an open issue. Indeed, the most powerful and popular AI tools —e.g., Neural Networks, Transformers, etc. — demand for heavy computational requirements. In this talk we will present the recent advancements in the resource-friendly AI realm, presenting various approaches for compressing AI, constructing more efficient models and integrate AI with symbolic components to ease their computational burden, focusing on possible cybersecurity entanglements.