Article 4 obligation requiring providers and deployers to ensure staff have sufficient understanding of the AI systems they work with, their capabilities, limitations and risks. Applies regardless of risk tier, in force since February 2025.