These rules need to be structured all-around and supported by focus parts, including explainability or fairness, close to which requirements might be developed and methods is usually aligned. Robustness: AI-powered devices need to be actively defended from adversarial attacks, minimizing prot