EU AI Act transparency rules are now in force: what Irish training providers should review

For training providers using artificial intelligence, 2 August 2026 was an important date. Transparency obligations under Article 50 of the EU AI Act began to apply across the EU, while Ireland has also moved into the implementation and enforcement phase.

For Irish course providers, this creates two separate but connected issues. First, providers should review how AI is being used inside their own training operations. Second, there is a growing need for credible professional education that helps organisations and employees understand how to use AI responsibly.

Neither means that every trainer needs to become an AI lawyer. But AI governance, transparency and literacy are increasingly becoming part of sound professional training practice.

1. Know where AI appears in your learner journey

Start with a simple inventory. You may already be using AI in more places than you realise:

• creating first drafts of course materials;
• generating images, video or narration;
• providing learner-facing chatbots or virtual tutors;
• generating quizzes or assessment questions;
• analysing learner feedback;
• personalising course recommendations;
• producing marketing content;
• summarising learner submissions;
• helping trainers prepare feedback.

These uses are not all treated identically under the AI Act.

Article 50 includes transparency requirements for certain directly interactive AI systems and for particular forms of AI-generated or manipulated content. That does not mean every PowerPoint slide drafted with an AI tool suddenly needs an “AI-generated” banner.

Context matters, as does the organisation's role in relation to the particular AI system. The practical first step is much simpler: document where AI is present, what it is doing, who uses it and who checks the output.

2. Make AI interactions clear to learners

Imagine an online course includes a conversational assistant that answers learners' questions.

If a learner could reasonably believe they are communicating with a person, transparency becomes important. The European Commission's Article 50 guidance addresses AI systems designed for genuine direct interaction with people.

From a course-design perspective, the solution can also be straightforward. Instead of presenting a tool simply as “Ask your tutor”, an appropriate interface might say “Ask our AI learning assistant”.

That is not just a compliance consideration. It is good learner experience. People should understand what sort of interaction they are having and calibrate how much reliance to place on the response.

3. Review synthetic video, audio and imagery

AI-generated presenters, synthetic voices and automated video production are becoming increasingly accessible to training companies.

These tools can dramatically reduce the time and cost involved in creating or updating eLearning. But providers should understand the distinction between ordinary AI-assisted content creation and content that may fall within the AI Act's rules concerning deepfakes or materially AI-generated or manipulated media.

The European Commission identifies deepfake disclosure as one of the areas covered by Article 50's transparency requirements.

A sensible content-production process should therefore record:

• which AI tools were used;
• which parts of the content were synthetically generated;
• whether a real person is being represented;
• whether disclosure may be required or appropriate;
• who reviewed and approved the final material.

The objective should not be to label content indiscriminately. It should be to avoid misleading learners about what they are seeing or hearing.

4. Keep meaningful human review in the course-production process

Generative AI is extremely useful for creating outlines, examples, questions, summaries and first drafts.

But professional course providers should avoid turning “AI generated it” into their quality-assurance process.

A competent person should remain responsible for reviewing:

• accuracy;
• relevance;
• learning outcomes;
• examples and scenarios;
• sources;
• assessment questions;
• sector terminology;
• potentially outdated material.

This becomes particularly important in courses covering areas such as HR, health and safety, finance, healthcare, compliance or regulation.

Regardless of whether a particular training document falls within a specific legal transparency provision, human editorial ownership is a strong quality standard for professional learning.

5. Don't overlook AI literacy inside your own organisation

The AI Act is not only concerned with labels and disclosures. Article 4 addresses AI literacy.

The obligation for providers and deployers of AI systems to take measures to support the AI literacy of staff has applied since February 2025, while national supervision and enforcement of the provision began in August 2026.

The provision does not prescribe one universal training course or one fixed level of competence for everyone. That makes practical sense: the AI literacy needed by someone using an AI tool to draft marketing copy is different from that required by someone introducing AI into recruitment, learner assessment or safety-critical decision-making.

For training providers, a useful internal programme might therefore cover:

• which AI tools the business permits;
• appropriate and inappropriate use;
• checking AI-generated information;
• confidentiality and data handling;
• bias and over-reliance;
• human oversight;
• transparency with learners;
• what to do when an AI output appears unreliable.

The strongest approach is role-based and practical: train people for the AI tools, responsibilities and risks they actually encounter.

6. This also creates a training opportunity

The AI Act creates an obvious new learning need for Irish employers. But training providers should resist the temptation to launch a generic “AI Act Compliance Certificate” after reading a few online summaries.

A stronger course starts with the learner's role.

AI literacy for employees

This might focus on safe everyday use, checking outputs, confidentiality, transparency and knowing when to escalate.

AI governance for managers

This could address organisational controls, accountability, risk, oversight and decision-making.

AI for HR teams

This might explore the particular issues around recruitment, employee information and automated decision support.

Responsible generative AI for marketing teams

This could concentrate on verification, intellectual property, transparency, approval processes and appropriate use.

This creates much more meaningful professional development than trying to compress the entire regulation into one universal course.

Where legal compliance is discussed, providers should also be explicit about the boundary between professional education and legal advice.

A practical checklist for course providers

1. Map your AI use. Identify where AI is used in course development, delivery, assessment and learner interaction.
2. Check learner transparency. Make sure learners understand when they are genuinely interacting with an AI system rather than a person.
3. Review synthetic content. Establish when generated video, audio or imagery may require additional disclosure or review.
4. Introduce human quality control. Give a competent person responsibility for approving AI-assisted professional content.
5. Train your own team. Make AI literacy appropriate to the tools, responsibilities and risks relevant to each role.

For detailed compliance questions, providers should use the European Commission's official AI Act resources, the AI Office of Ireland and appropriate legal or regulatory advice.

The role of good professional training is different: translating changing expectations into knowledge and behaviours people can actually apply. And that is exactly where training providers can add value.

Sources and further reading

• European Commission — Guidelines on transparency obligations for providers and deployers of AI systems
• European Commission — Transparency obligations under Article 50 of the AI Act: Questions & Answers
• European Commission — AI Literacy: Questions & Answers
• European Commission — AI talent, skills and literacy
• Department of Enterprise, Tourism and Employment — AI Office of Ireland established, 30 July 2026