The initial proposal for an EU act regulating Artificial Intelligence, commonly referred to as the AI Act, has undergone substantial modifications under the Slovenian, French and Czech presidencies since it’s proposal by the Commission in spring 2021. As the content and details change over time it has become necessary to analyse to what extent the act will build trust towards AI as technology. In this position paper the ambition is to highlight one aspect of the act – the relation to trust – based on the version drafted for the general approach by the Council. Now, even reducing the analysis to one aspect is a complex task and the chosen points are the ones we have seen gaining most interest with the stakeholders we have interacted with. There are other points that could have been included but we have chosen to leave them out so as to not obscure the overall picture by all the details. The text does not follow a traditional structure of Introduction – Method – Results. Rather it represents nuggets we have dug out while exploring the proposed regulation, here analysed in relation to the notion of trust. We also assume that the reader has a basic understanding of the initial proposal. For those interested in a more detailed account of our method we recommend our contribution on agile principles for analysing upcoming regulations. Before we get going, we want to make it clear that while we are indebted to numerous representatives from commercial enterprises, public administrators spanning across EU departments to Swedish municipalities as well as researchers from different institutions, we are solely responsible for the claims and the analysis presented in this contribution.