OmnAI lab (pronounced /ɒmn/ɑɪ/) is a research lab 🔬based in Sapienza, University of Rome. Omnia is Latin for "everything" and AI stands for "artificial intelligence"🤖; the Om 🕉 is also the original vibration of the universe. We study computational models of learning and inference and, more generally, study how to make machines intelligent 🧠 to a certain degree (Artificial Intelligence). We believe that, despite recent progress, the way towards Artificial General Intelligence (AGI) is far away and we need inputs from multidisciplinary fields to make this happen. We focus our research mainly in the visual domain (computer vision)🌁 but we also draw inspiration from other fields such as computer graphics, natural language processing 📚. Our research is mainly empirically centered on machine learning and deep learning yet their foundations rest on mathematical tools 🧮 like linear algebra, high-dimensional geometry, and optimization. We currently have several lines of research that are all intertwined, touching on multiple aspects: adversarial robustness, proactive defense to image manipulation, inverse problems, and generative AI.