Examining a structural challenge in the EU AI Act arising from the increasing use of orchestration frameworks in agentic AI systems.The piece explains how these frameworks do not fit within the AI Act’s binary distinction between ‘providers’ and ‘deployers’. It outlines why this creates a governance gap, particularly in relation to accountability, conformity assessment, and cross-border data and tool use, and discusses the limits of existing mechanisms, such as Article 25 of the AI Act, in addressing this issue.
Lloyd Alexander Jones (Thu,) studied this question.