EN
English
ES
Spanish
On Sunday, 2 August 2026, national market surveillance authorities across the European Union gain formal supervisory powers to enforce Article 4 of the EU AI Act. The obligation itself has been in force since 2 February 2025, but this weekend marks the moment the enforcement infrastructure becomes fully operational, in Spain through AESIA and in every other Member State through its designated authority.
This article explains what Article 4 requires, who it covers, what actually changes on Sunday, and how the recent Digital Omnibus reshaped the wording of the duty.
What Article 4 requires
Article 4 of Regulation (EU) 2024/1689 obliges providers and deployers of AI systems to take measures aimed at supporting a sufficient level of AI literacy among their staff and any other persons operating AI systems on their behalf.
Following the Digital Omnibus, endorsed by the European Parliament on 16 June 2026 and approved by the Council on 29 June 2026, the standard shifted from an obligation to ensure a sufficient level of AI literacy to an obligation to support its development. In legal terms, this is a move from an obligation of result to an obligation of effort. The underlying duty remains, and the enforcement date has not moved.
Who the obligation covers
Article 4 applies to every organization that provides or deploys AI systems within the European Union, regardless of size, sector, or the risk classification of the system in question. The obligation reaches:
- Staff who directly operate AI systems
- Contractors and third parties acting on the organization’s behalf
- Employees whose work is meaningfully affected by the outputs of AI systems, even where they do not operate them directly
A common misreading is that Article 4 applies only to technical staff, such as engineers, data scientists, or developers. It does not. Any person whose role touches AI systems falls in scope.
The proportionality standard
Article 4 does not prescribe a fixed curriculum, a certification, or an examination. Instead, it requires literacy proportionate to three factors:
- The context in which the AI system is used
- The technical complexity of the system
- The role of the person concerned, together with the persons or groups on whom the AI system is used
Where an AI system affects vulnerable groups, for example job applicants, elderly customers, or recipients of social benefits, the literacy standard for the people operating the system is correspondingly higher.
What changes on Sunday, 2 August 2026
Three things become operational on this date:
- National market surveillance authorities gain formal supervisory powers over Article 4. In Spain this is AESIA. Other Member States have designated equivalent bodies, including BNetzA in Germany and the newly established National AI Office in Ireland.
- The broader enforcement architecture of the AI Act comes into force, including cross-border coordination between national authorities.
- Most of the AI Act’s remaining obligations become applicable. The notable exception is the set of high-risk system obligations, which the Digital Omnibus deferred to 2 December 2027 for standalone Annex III systems and to 2 August 2028 for AI in regulated products under Annex I.
What the Digital Omnibus did, and did not, change
The Digital Omnibus deferred the applicability of high-risk system obligations under Annex III and Annex I. It did not defer Article 4. It softened the standard from ensuring a sufficient level of literacy to supporting its development, but the duty, the applicability dates, and the 2 August 2026 enforcement start remain in place.
For organizations whose AI systems are not high-risk, Article 4 is the only EU AI Act obligation that becomes enforceable in the near term. For organizations that operate high-risk systems, it is the practical first step ahead of the December 2027 documentation deadline.
Legal basis
- Regulation (EU) 2024/1689 (EU AI Act), Article 4
- Regulation (EU) 2024/1689, Article 99 on penalties
- Digital Omnibus, adopted by the Council on 29 June 2026
