Late last year, much of the coverage around the EU AI Act focused on the decision to delay enforcement of high-risk systems. What received far less attention was a structural change that may prove more consequential for organisations in practice: the move from oversight by national authorities to a self-assessment model for compliance. In effect, responsibility for determining whether an AI system meets the Act’s requirements now sits squarely with the organisation deploying it. This shift turns the delay into something closer to a working period than a pause. It is a chance to put governance, literacy, documentation, and risk management on a more deliberate footing, rather than reacting under deadline pressure. Read the full article here 👉 on our website #euaiact #aicompliance #aistrategy #aipolicy