Best practice, currently under review
Applies since February 2025. Article 4 requires providers and deployers of AI systems to take measures, to their best extent, to ensure a sufficient level of AI literacy among staff and other persons dealing with the operation and use of AI systems on their behalf. The Digital Omnibus proposal (19 Nov 2025) may shift this obligation toward Member States rather than individual organisations.
Firm requirement - unaffected by Omnibus
Article 4 · General AI literacy
Article 26 · Human oversight
Staff operating high-risk AI systems (HR-tech, scoring, biometrics) must be trained for effective human oversight. This obligation remains in place regardless of Article 4 amendments - it requires deep tailoring to the specific high-risk system in use.