Skip to main content

news_analysis

EU AI Act AI Literacy Requirements for Managers: A Softer Rule Meets Live Enforcement

By AInews Desk
EU AI Act AI Literacy Requirements for Managers: A Softer Rule Meets Live Enforcement
7 min read
1240 words

The two changes landed a few weeks apart. The Digital Omnibus amendments to Article 4 of the AI Act entered into force in mid-July 2026, and the supervision and enforcement rules apply from 2 August 2026 onwards. For managers, the EU AI Act's AI literacy requirements in 2026 reflect both of these developments: the amendments to the wording of the duty and the start of enforcement by national authorities.

This piece covers what Article 4 now asks of a company whose staff use tools like ChatGPT for writing or translation tasks, what it does not ask, and where the obligation is still strict.

The duty has applied since February 2025

Under the AI Act, the AI literacy obligation already applies before 2026. Article 4 entered into application on 2 February 2025, so, in the Commission's words, the obligation to take measures supporting staff AI literacy "already applies". What changed this year is the text of the obligation and the start of enforcement.

The Commission presented the Digital Omnibus on 19 November 2025 as part of a wider effort to simplify the EU's digital rulebook. On Article 4, it proposed shifting the obligation onto the Member States and the Commission to promote AI literacy and upskilling, "rather than enforcing an unspecific obligation on organisations".

That is not the version that became law. Companies still carry a duty. The amended Article 4(1), as shown on the AI Act Service Desk, says providers and deployers "shall take measures to support the development of AI literacy" for their staff and for others operating AI systems on their behalf. Those measures should take into account the people's technical knowledge, experience, education and training, the context of use, and who the systems are used on. The amended paragraph also adds this sentence: "This obligation does not require providers or deployers to guarantee any specific level of AI literacy of any individual."

In practice, the law now asks for effort, not results. Nobody can fail you because one employee still doesn't understand how a language model works.

Who enforces Article 4, and starting when?

Enforcement sits with national authorities, not with Brussels. The Q&A says supervision and enforcement of Article 4 is not with the AI Office, but under the remit of national market surveillance authorities, which start supervising and enforcing the rules as of 2 August 2026. According to the same Q&A, national market surveillance authorities could impose penalties and other enforcement measures for Article 4 infringements, based on national laws Member States were due to adopt by 2 August 2025. The Service Desk's version of the Act is based on the EUR-Lex consolidated text as at 27 July 2026, which includes the Omnibus changes.

DateEventImpact
2 Feb 2025Article 4 enters into applicationLiteracy duty applies
2 Aug 2025National penalty laws dueBasis for national sanctions
19 Nov 2025Digital Omnibus presentedArticle 4 rewrite proposed
Mid-July 2026Omnibus amendments in forceNo guaranteed literacy level
27 Jul 2026EUR-Lex consolidated textService Desk text updated
2 Aug 2026Supervision and enforcement applyAuthorities can act

The Commission's Q&A says enforcement "follows a proportionate approach". Sanctions must reflect the individual case, the nature and gravity of the infringement, and whether it was intentional or negligent. One warning in the Q&A deserves attention: a sanction "might, however, be more likely if there is proof of an incident due to lack of appropriate training and guidance of employees or other persons". So the practical risk is an incident followed by an investigation that finds no training was ever given.

What you can skip

The Commission's answers are direct on these points:

  • No certificate. There is no need for a certificate.
  • No AI officer or governance board. No specific governance structure is mandated to comply with Article 4.
  • No knowledge testing. Article 4 does not entail an obligation to measure the knowledge of AI of employees.
  • Simple records are enough. Organisations can keep an internal record of trainings and/or other guiding initiatives.

The obligation also reaches beyond your payroll. For contractors and service providers acting on your behalf, the Q&A says it depends on the type of AI system and the risk, but in general they "need to have the appropriate AI skills to fulfil the task in question (same as the employees)".

Help is coming from the public side as well. Under amended Article 4(2), the Commission shall publish practical examples of how to comply on the single information platform, with particular support for SMEs. Under Article 4(3), the Board shall adopt recommendations that set out common objectives.

The ChatGPT case managers actually face

The Commission's Q&A covers the most common workplace situation directly. It asks whether a company whose employees use ChatGPT for writing advertisement text or translating text has to comply with Article 4. The Commission's answer is short: "Yes, they should be informed about the specific risks, for example hallucination."

For a manager, that usage is the starting point. The training for staff using a chat assistant should take into account their technical knowledge, experience, education and training, as well as the context the AI systems are to be used in. In practice, staff should know that these tools can produce fluent text that is wrong (often called hallucination), that a machine translation needs checking before it goes out, and what information should never be pasted into a prompt. Whatever format you choose, keep a dated record of it. The Q&A also warns that simply relying on the AI systems' instructions for use or asking the staff to read them might be ineffective in many cases.

Where the obligation stays strict

The Omnibus softening has a clear limit. The Commission states that for those who deploy high-risk AI systems, the obligation to ensure that their staff is trained to ensure human oversight remains in place.

That changes things for HR and operations leaders. If your organisation runs an AI system that falls into a high-risk category, a general obligation to support the development of AI literacy is not enough. The people responsible for human oversight need to actually be trained for that role: the Q&A points to Article 26, which introduces an obligation for deployers of high-risk systems to ensure that the staff dealing with the AI systems in practice is sufficiently trained to handle the system and ensure human oversight. Treating both cases the same, with a light general briefing for ChatGPT users and the same briefing for a high-risk deployment, is the mistake to avoid.

What managers should do now

For most managers, the requirements are light, but they are enforceable. Make a reasonable, documented effort that fits how your staff actually use AI. Don't buy certificates or set up new governance roles you don't need. Keep a record of what you did. Then check separately whether any system you deploy is high-risk, because that is where a general awareness course stops being enough. Since 2 August 2026, the national authorities that can enforce these rules have been in a position to ask what you have done.

Sources

7 min read
•
1240 words

Tags

EU AI Act AI literacy requirements for managersArticle 4 EU AI Act manager training obligationsAI literacy compliance Article 4 requirementsEU AI Act manager responsibilities

Related Articles