ELLE: European Doctoral Network on Secure and Safe Artificial Intelligence in the Age of Foundation Models

ELLE proposes a European doctoral network aimed to train a new generation of AI scientists and professionals that will promote and reinforce the European leadership in the development of secure and safe AI technology. Within ELLE, we will tackle four main scientific and technical challenges that are hindering the concrete implementation of legally-compliant AI models with the provisions of the EU AI Act and related regulations, especially after the recent paradigm shift that has fostered the wide adoption of foundation models. These challenges include ensuring the main trustworthiness properties of learning algorithms, i.e., (i) technical robustness and safety; (ii) privacy and data governance; and (iii) interpretability and transparency; while considering models that can be (iv) continuously adapted in security-sensitive and safety-critical tasks where the data is streaming, non-stationary and showing unexpected changes of its statistical characteristics. ELLE aims to tackle these challenges jointly, building upon three main pillars: (1) the proposal of 15 distinct, multi-disciplinary Ph.D. research programs that are deeply intertwined within the four aforementioned challenges via carefully defined co-supervisions and secondments; (2) the development of open courses, educational material, and workshops, with an emphasis on hands-on, collaborative learning and attention to transferable skills; and (3) the promotion of mobility and knowledge transfer; all within a training network that includes a selection of Europe's finest researchers, supported by the ELLIS Society and stemming from the ELSA European Network of Excellence. ELLE will integrate novel and rigorous science developments with training skilled doctoral researchers to nurture and sustain the development and deployment of AI technology in compliance with the EU AI Act and related regulations, serving and promoting foundational EU values.