The demand for automated part-inspection using CT inspection is constantly increasing. So far, available commercial solutions for data evaluation and defect detection in CT data are reaching their limitations and none is Airbus qualified. Therefore, all qualified evaluations still have to be done manually according to a specific inspection process. As a consequence, the evaluation time is high and overall inspection costs contribute significantly to the costs of the additive manufactured parts. To overcome these limitations, Airbus Central R&T has been working on AI based algorithmic methodologies to support shop floor inspectors in this process. Airbus Central R&T has established a cross-domain project consisting of data scientists (AI experts) and NDT/CT experts to better tackle the challenge for higher degrees of automation supported by AI. The goal of the project was to develop AI based algorithmic concepts for CT defect detection and to integrate them into a proof-of-concept demonstrator. In the first project phase, seeded defect types for automated detection have been specified and corresponding CT data-sets with artificial defects were selected. Various algorithmic methodologies have been investigated. In the second project phase, the most promising approaches have been identified and integrated into the proof-of-concept demonstrator software. This paper will discuss the potential AM defect types considered in the project, the available data samples with seeded defects as well as the high-level algorithmic approach and the demonstrator we have developed. The final section will conclude with an outlook which is motivated by the promising initial results and the need for a qualifiable industrial process solution based on EASA level 1 requirements for AI utilization.