This tutorial explains, in the form of slides with audio, the proposal for an EU AI Act, as proposed by the European Commission in the Spring of 2021. It does not discuss the subsequently proposed amendments.

Key issues discussed are: (1) the overall architecture of the AI, (2) the pragmatic approach to the definition of AI systems (which is not about ‘AI’ but about ‘AI systems’), (3) the different roles, notably that of the providers of these systems, (4) the emphasis on high risk AI systems and (5) the details of the requirement that must be met by all high risk systems. It also explain what AI practices are prohibited and what transparency requirements must be met by a small set of AI systems.