AI literacy training in Ireland: what the EU AI Act asks of marketing teams
If your team uses ChatGPT, Claude or Copilot at work, the AI Act asks you to help them use it well. Here is what that means, what changed in July 2026, and what we would do.
If your marketing team uses ChatGPT, Claude, Copilot or any other AI tool at work, the EU AI Act asks your company to take steps to help those people use it well. That's Article 4, the AI literacy rule. It has applied since 2 February 2025, it was softened in July 2026, and it still applies to you. You don't need a certificate, there is no set course, and no fine is named for it in the Act. But you do need to be able to show you did something sensible.
Here is the plain version, from the text of the law and the European Commission's own guidance, with links at the bottom.
Does the AI Act apply to a marketing team using ChatGPT?
Yes. The Act puts the AI literacy duty on "providers and deployers" of AI systems. A provider builds the tool. A deployer is any organisation that uses one at work. So a company whose marketers write ad copy with ChatGPT is a deployer.
The Commission's AI literacy Q&A uses almost exactly that example. It asks whether a company whose employees use ChatGPT for "writing advertisement text or translating text" needs to comply, and answers yes, saying staff "should be informed about the specific risks, for example hallucination." (Hallucination is the polite word for the tool making things up with total confidence. We've all met someone like that.)
It isn't only your employees, either. The rule covers "other persons dealing with the operation and use of AI systems on their behalf", and the Commission's Q&A names contractors, service providers and clients as examples. So your freelancers and your agency count too.
What changed in July 2026?
The rule got softer. The original Article 4 said companies "shall take measures to ensure, to their best extent, a sufficient level of AI literacy" of their staff. That was vague, and nobody could tell you what "sufficient" meant.
In July 2026 the EU passed the Digital Omnibus on AI, Regulation (EU) 2026/1744, which replaced Article 4. It was published in the Official Journal on 24 July 2026 and came into force three days later. The new wording says providers and deployers "shall take measures to support the development of AI literacy" of their staff, and adds that this "does not require providers or deployers to guarantee any specific level of AI literacy of any individual."
So the duty moved from making sure people are good enough to helping them get better. The new text also tells the Commission and EU countries to support companies with this, "in particular SMEs", and asks the Commission to publish practical examples.
Is there an exemption for small businesses?
Not in Article 4. The rule has no size limit. What the new text gives small and medium firms is help, not an exemption. If you're a team of three using Canva's AI tools, the rule still applies, but nobody expects a three-person team to run a training academy. What counts as sensible depends on the tools you use and what you use them for.
Do we need a certificate or a formal course?
No. The Commission's Q&A says "There is no need for a certificate" and that "no strict requirements or mandatory trainings are imposed." There's also no duty to test people.
It does say "Organisations can keep an internal record of trainings and/or other guiding initiatives." That's worded as a choice, not a rule. But if anyone ever asks what you did, a record is the only thing that answers them, so I'd keep one.
Can we be fined?
The fine tiers in Article 99 of the Act don't list Article 4. They list things like banned practices, high-risk systems and the transparency rules. But Article 99 also tells each EU country to set penalties for "any infringement of this Regulation", and the Commission's Q&A says national market surveillance authorities "could impose penalties and other enforcement measures" for breaches of Article 4. It adds that any sanction must be proportionate, and that one "might, however, be more likely if there is proof of an incident due to lack of appropriate training."
Supervision and enforcement by national authorities started on 2 August 2026.
My read, and it is a read rather than legal advice: a team that has done nothing is the one at risk, and the risk turns real when something goes wrong. An AI-written ad that makes a false claim, or a customer list pasted into a free chatbot, is the kind of incident where "we never told anyone anything" looks bad. Since July, "we took reasonable steps to help our people" is a much easier thing to show than "everyone reached a sufficient level."
Who enforces the AI Act in Ireland?
Ireland has chosen what the government calls "a distributed model of competent authorities." The AI Office of Ireland was set up in summer 2026 as the central body that coordinates the Act's implementation. The supervising itself is shared between existing regulators, including the Central Bank, the Data Protection Commission and the Competition and Consumer Protection Commission, each for certain kinds of AI system. The AI Office publishes the list of national competent authorities on its website.
What I couldn't find, at the time of writing, is a public statement of which authority handles Article 4 for an ordinary company using everyday AI tools. If that matters to you, ask the AI Office directly.
What should a marketing team actually do?
This is what I'd do, and it's my advice rather than anything the Act spells out.
Start by writing down which AI tools your team and your agencies use, and what for. Count the AI built into tools you already pay for, like Canva, HubSpot, Google Ads and Meta, not just the chatbots.
Then write one page on the risks that matter for marketing work. Tools make things up, so every fact and claim gets checked by a person before it goes out. Customer data and anything confidential never goes into a tool your company hasn't approved. And Meta and Google can now change your ads with their own AI features, so someone checks what actually ran.
Next, take everyone who uses these tools through that page, freelancers and agencies included, and show them how to use the tools properly for the work they actually do. Keep a simple record of who did what and when. Do it again when your tools change, which right now is roughly every Tuesday.
None of that is expensive. Done well, it also makes your team better at the job, and that was the point of the rule all along.
If you'd like help with the second half, our free course, AI Fluency for Ambitious Marketers, teaches marketers to use these tools well for real work. It's training, not a compliance certificate, and the Act doesn't ask for one. The link is at the foot of this page.
This note was drafted with AI and checked by Paul Dervan before it was published.
Sources
The original Article 4, in Regulation (EU) 2024/1689 as published in the Official Journal on 12 July 2024: eur-lex.europa.eu. The Digital Omnibus on AI, Regulation (EU) 2026/1744 of 8 July 2026, which replaced it: eur-lex.europa.eu. The European Commission's AI literacy Q&A, last updated 20 October 2026: digital-strategy.ec.europa.eu. The Department of Enterprise's page on the General Scheme of the Regulation of Artificial Intelligence Bill 2026: enterprise.gov.ie. The AI Office of Ireland and its list of national competent authorities: aioffice.gov.ie.
This is a plain-English guide, not legal advice. If you build AI systems, or use AI for anything the Act calls high-risk (hiring decisions or credit scoring, for example), more of the Act applies to you than this piece covers.