Brain-Computer Interfaces are communication systems that bypass the natural nerve-muscle communication pathway and directly exploit information from the brain.These are very useful systems for people that have a damaged nerve-muscle communication, as they offer new opportunities to interact with the environment via brain commands sent to external mechanical devices. However, most progress is currently made in invasive BCI which requires surgeries. Recording brain signals non-invasively is challenging but there are constantly new technical advances allowing for more sophisticated and multimodal measures.My aim is to widen and improve the number of applications of non-invasive brain computer interfaces through the application of artificial intelligence (AI). These methods enable the detection of very complex temporal and dynamic features of non-invasively recorded brain signals. AI also improve the fusion of signals with very different physical and physiological origins. I believe the personalisation of AI algorithms and the advances in multimodal brain signal recording technologies will enable people with tetraplegia and similar disabilities regain autonomy in their daily life and improve the diagnostic and rehabilitation of neurological conditions.