As the impact of artificial intelligence (AI) can include benefit and harm, the AI Act (AIA) takes a risk-based approach, putting an emphasis on the development of secure, trustworthy and ethical AI. In terms of the various categories of the AIA, EU values play an important role in ‘prohibited’ AI, fundamental rights in ‘high-risk’ AI, and soft-law (and ethical principles) in ‘low-risk’ AI, referring to voluntary ‘codes of conduct’. Relying on values is an obvious approach in the EU, being a ‘Union of values’ since the Lisbon Treaty. While the hierarchy of EU law results in the legal obligation for EU secondary law to be in line with EU primary law (i.e., the values enshrined in Art 2 TEU), it is convincing in a particularly sensitive and impactful field, as in the case of AI, to put a special emphasis on values and fundamental rights. Values rather occur in the preamble, hardly in the binding text of the AIA. An exception is the keyword of ‘fundamental rights’ that occurs 44 times in the articles, due to the importance for ‘high-risk’ AI, followed by other values and principles. Some EU values do not occur at all (e.g., pluralism and tolerance). Rather abstract values can be combined with more concrete legal principles, where the latter are legally binding, have legal addressees and can have legal consequences. Hence, it is not very surprising that the principle (and value) of non-discrimination plays an important role in the AIA. Consequently, the overall-approach of the AIA can be summarised as ‘human-centric’, hence related to ‘human well-being’ and based on the ideas of ‘humanism’.