Resources, EU AI Act
AI literacy under the EU AI Act: Article 4 after the Omnibus
Every company that uses AI systems has an AI literacy duty under Article 4 of the EU AI Act. This guide explains the original rule, what the Digital Omnibus changed, and how to build and record a proportionate programme.
Updated 20 November 2026. 8 minute read.
In short
- Article 4 has applied since 2 February 2025 to all providers and deployers of AI systems, whatever their size and whatever the risk level of the system.
- The original text required measures to ensure, to the best extent, a sufficient level of AI literacy. The Digital Omnibus on AI, Regulation (EU) 2026/1744, replaced Article 4 with effect from 27 July 2026: companies must now take measures to support AI literacy, and no specific level is required for any individual.
- The Commission's AI literacy Q&A says no certificate is needed and that companies can keep an internal record of training and other initiatives.
- A proportionate programme is role-based. Employees who approve or oversee agents need more than general awareness: what the agent does, its limits, how to check its work and when to stop it.
What Article 4 originally required
The AI Act, Regulation (EU) 2024/1689, entered into force on 1 August 2024. Article 4 on AI literacy was among the first rules to apply, from 2 February 2025. As adopted in 2024, it read:
"Providers and deployers of AI systems shall take measures to ensure, to their best extent, a sufficient level of AI literacy of their staff and other persons dealing with the operation and use of AI systems on their behalf, taking into account their technical knowledge, experience, education and training and the context the AI systems are to be used in, and considering the persons or groups of persons on whom the AI systems are to be used."
Article 3(56) defines AI literacy as the skills, knowledge and understanding that allow providers, deployers and affected individuals to make an informed deployment of AI systems, and to gain awareness of the opportunities and risks of AI and the harm it can cause.
Three points in the original wording shaped how companies read it. The duty was to ensure a sufficient level, which suggested an outcome to reach. It was qualified by to their best extent. And it had to be tailored to the staff concerned, the context of use and the individuals affected by the system.
How the Digital Omnibus changed Article 4
The Digital Omnibus on AI, Regulation (EU) 2026/1744 of 8 July 2026, amended the AI Act. The Council gave its final approval on 29 June 2026. The regulation was published in the Official Journal on 24 July 2026 and entered into force on 27 July 2026. It replaced Article 4 in full.
What changed
- Providers and deployers must now take measures to support the development of AI literacy among their staff and others who operate or use AI systems on their behalf, instead of measures to ensure a sufficient level.
- The new text makes clear that the duty does not require a specific level of AI literacy to be guaranteed for any individual.
- A new paragraph gives the Commission and the Member States a role in promoting AI literacy and supporting providers and deployers, in particular SMEs.
What did not change
The duty still sits with companies. The Commission's November 2025 proposal would have moved the main responsibility for AI literacy to the Commission and the Member States. The co-legislators did not accept that in full and kept an obligation on providers and deployers. The duty still covers all AI systems, not only high-risk ones, and it is not delayed. The Omnibus changes to the high-risk timeline do not affect it.
Article 4 is not on the list of specific fines in Article 99. Article 99(1) leaves Member States to set penalties for other breaches of the Act, so national rules decide the consequences. Check the rules where you operate.
The wording above summarises the amended Article 4. Read the consolidated text of the AI Act on EUR-Lex for the exact legal wording before relying on it in a policy.
Who the duty applies to
Article 4 applies to every provider and every deployer of an AI system. A deployer uses an AI system under its authority in a professional activity (Article 3(4)). A company whose employees use AI agents to draft emails, triage tickets or update records is a deployer, and so it has the duty.
- Size does not matter. Start-ups and SMEs are covered, although the Commission and Member States must support SMEs.
- Risk level does not matter. The duty applies to a simple writing assistant as much as to a high-risk system.
- It covers staff and others who operate or use AI systems on the company's behalf. The Commission's Q&A gives contractors, service providers and clients as examples of others.
- Purely personal, non-professional use is outside the AI Act (Article 2(10)).
For high-risk systems, the Act adds a stricter, separate requirement. Article 26(2) requires deployers to assign human oversight to employees with the necessary competence, training and authority. Article 4 is the baseline for every provider and deployer; Article 26(2) is the higher bar for oversight of high-risk systems.
What the Commission's Q&A and guidance say
The Commission's AI Office published an AI literacy questions and answers page in 2025, together with a living repository of AI literacy practices from companies and public bodies. The Q&A was written for the original text, so read its references to a sufficient level in light of the Omnibus. Its practical points still help:
- There is no single approach for all companies and no mandatory training format. The AI Office signals flexibility.
- As a minimum, a company should ensure a general understanding of AI in the organisation, consider whether it is a provider or a deployer, and consider the risk of the AI systems it provides or uses.
- Relying only on a system's instructions for use, or asking staff to read them, may be ineffective and insufficient.
- No certificate is needed. A company can keep an internal record of training and other guiding initiatives.
- Copying a practice from the living repository does not by itself give a presumption of compliance.
- National market surveillance authorities supervise and enforce the rule, with supervision and enforcement rules applying from 2 August 2026.
The repository was revamped in November 2025. Use it for ideas on format and content, not as a template that guarantees compliance.
A proportionate programme for employees who oversee agents
Proportionate means matching training to the role and the risk. AI agents act rather than only suggest, so employees who approve or supervise agent actions need more than general awareness. A practical programme has three tiers.
All employees who use AI
- What AI agents are, what the company uses them for and which uses are not allowed.
- That agents can be wrong, confident and persuasive at the same time.
- What data may and may not be shared with AI tools.
- How to report a problem.
Employees who approve or oversee agents
- The agent's role, the apps and tools it can use, and its budget.
- How to check its work before approving: sources, figures, recipients and tone.
- Common failure modes, such as made-up facts and prompt injection hidden in emails or documents.
- Automation bias: the tendency to approve by habit. Approvals should be real decisions.
- How and when to pause or stop an agent, and whom to tell.
Agent owners and managers
- Setting roles, permissions, approval rules and budgets.
- The company's AI policy, the EU AI Act use-case register and data protection duties.
- Reading the activity log and reviewing agents on a schedule.
Keep it short and practical. A one-hour session per tier, a written guide and a review when an agent's role changes will usually do more than a long generic course.
How to document AI literacy
Documentation is not a separate legal requirement in Article 4, but a record is the simplest way to show the measures you took. The Commission's Q&A confirms that an internal record is enough. Keep:
- A short AI literacy policy: scope, tiers, owners and review cycle.
- A role matrix: which roles use, oversee or own which agents, and which tier of training each needs.
- The training materials, with a version date.
- A log of who completed which session and when.
- Records of other measures, such as written guides, briefings on new agents and lessons from incidents.
- A review date, and a trigger to update training when agents, models or uses change.
Link each agent in your use-case register to the employees who oversee it and the date they were trained. That one link answers most audit questions.
How Blentera helps
Blentera is a platform where companies run AI agents with employees in charge. Each agent has a role in the org chart, so it is clear which employees oversee it. Approval rules make oversight a concrete task, and each agent can use only the apps and tools it is given. Budgets warn at 80% and pause the agent at the limit. The EU AI Act use-case register records each use case, and the activity log, which can be exported, shows what agents did and who approved it. It is hosted in the EU. Blentera does not deliver AI literacy training, and meeting Article 4 remains your company's responsibility.
Questions
- Article 4 requires providers and deployers to take measures to support AI literacy. It does not prescribe a training format. Training is the most common measure, but guides, briefings and other initiatives can count too.
- Regulation (EU) 2026/1744 replaced Article 4 from 27 July 2026. Companies must now take measures to support AI literacy rather than ensure a sufficient level, no specific level is required for any individual, and the Commission and Member States must promote AI literacy.
- No. The Commission's AI literacy Q&A says no certificate is needed and that companies can keep an internal record of training and other initiatives.
- Yes. A company that uses AI systems in its business is a deployer, and Article 4 applies to all deployers, whatever their size or the risk level of the system.
- Article 4 has applied since 2 February 2025. According to the Commission's Q&A, national market surveillance authorities supervise and enforce it, with those rules applying from 2 August 2026.
Sources
- Regulation (EU) 2024/1689 (AI Act), original text, EUR-Lex
- Regulation (EU) 2024/1689 (AI Act), consolidated text of 27 July 2026, EUR-Lex
- Regulation (EU) 2026/1744 (Digital Omnibus on AI), EUR-Lex
- AI literacy: questions and answers, European Commission
- Living repository to foster learning and exchange on AI literacy, European Commission
- AI talent, skills and literacy, European Commission
- Council gives final green light to simplify and streamline AI rules, Council of the EU, 29 June 2026
Related
This guide is general information, not legal advice. Check how the rules apply to your company with your own advisers.
Get started
Put your first agent to work in minutes.
Three steps, and your first agent is working for your company.
Step 1
Create an account
Sign up with your work email and invite your team. No sales call needed.
Step 2
Add your first agent
Create an agent, give it a role, and connect your own AI key and the apps it may use.
Step 3
Give it a task
Ask in Blentera or your chat app, and follow the work until it is done.