Ireland is enforcing the EU AI Act

Jessica Bryan
•
8
min read
eu-ai-act-ireland
Copied

The EU AI Act now has real enforcement behind it in Ireland. On 31 July 2026, the Regulation of Artificial Intelligence Act 2026 took effect. It gives Irish authorities the power to investigate and fine organisations under the EU AI Act. If your business uses AI to hire, lend, serve customers or run operations, this applies to you.

What changed in July

Three developments took place within ten days:

  • 21 July: the Regulation of Artificial Intelligence Act 2026 was signed into law.
  • 27 July: an amendment to the EU AI Act came into force, postponing some deadlines for high-risk AI.
  • 30 July: the AI Office of Ireland was established, with Paul Byrne appointed as chief executive.

Ireland has adopted a distributed model of enforcement. Rather than creating a single AI regulator, it has given responsibility to existing sector regulators, with the AI Office of Ireland coordinating their work. For most organisations, this means the regulator they already report to will also oversee their use of AI.

What has been delayed, and what hasn't

The requirements for high-risk AI systems have been postponed. AI used in recruitment, education, credit scoring and critical infrastructure must now comply by 2 December 2027. AI built into products already covered by EU safety legislation, such as medical devices, has until August 2028.

The remaining obligations are unchanged. Prohibited AI practices have been banned since February 2025, and two further prohibitions take effect on 2 December 2026. Since August 2026, organisations have been required to tell people when they are interacting with an AI system and to label AI-generated content such as deepfakes.

AI literacy is already a legal requirement

Article 4 of the EU AI Act requires organisations that develop or use AI to take measures to build AI literacy among their staff and anyone operating AI on their behalf. This obligation has applied since February 2025 and was not affected by the July postponement.

It is also one of the more straightforward requirements for a regulator to assess. Rather than reviewing technical systems, a regulator can ask an organisation to demonstrate:

  • where AI is used across the organisation, and by whom
  • what those people understand about the tools, including their limitations
  • how decisions informed by AI are reviewed

A written AI policy is a useful starting point, but on its own it is unlikely to be sufficient evidence.

Why this matters for leadership teams

  • Cost: addressing AI literacy now is considerably less expensive than responding to it under regulatory scrutiny.
  • Risk: penalties for the most serious breaches can reach 7% of global annual turnover, and Irish authorities now have the power to impose them.
  • Competitive position: organisations that can demonstrate responsible use of AI are better placed in procurement, partnerships and due diligence.

How Data & AI Literacy Academy can help

Data & AI Literacy Academy works with organisations to build the data and AI literacy the EU AI Act requires, from board level to the teams using AI day to day. Our programmes give leaders a clear account of how AI is governed in their organisation, and give employees the knowledge to use it responsibly and effectively.

FAQ

Does the EU AI Act apply to my organisation in Ireland?

If you use AI in the EU, it very likely does. The Act covers organisations that build AI and organisations that use it. That includes off-the-shelf tools.

Does the EU AI Act apply to companies outside the EU?

Yes. If you sell AI into the EU, or the output of your AI is used in the EU, the Act can apply to you. That includes UK and US companies working with Irish clients.

When did the EU AI Act come into force in Ireland?

The EU AI Act entered into force across the EU on 1 August 2024, with its rules phasing in over several years. Ireland's own enforcement law, the Regulation of Artificial Intelligence Act 2026, took effect on 31 July 2026.

Who enforces the AI Act in Ireland?

Existing sector regulators, coordinated by the AI Office of Ireland. Which regulator you deal with depends on your sector.

Weren't the high-risk rules delayed?

Yes. Most high-risk AI now has until 2 December 2027, and AI built into regulated products has until August 2028. Banned practices, transparency duties and AI literacy were not delayed.

What counts as high-risk AI?

AI used in areas that can seriously affect people's lives. That includes recruitment and HR, education, credit scoring and critical infrastructure. It also covers AI that acts as a safety component in regulated products, such as medical devices.

What are the penalties under the EU AI Act?

Using a banned AI practice can cost up to €35 million or 7% of global annual turnover, whichever is higher. Other breaches carry lower caps, and smaller businesses face reduced maximums. In Ireland, authorities can now investigate and impose these penalties.

What does the AI literacy rule ask for?

It asks you to take steps so your staff, and anyone using AI on your behalf, understand the AI they work with. That should reflect their role, their experience and how the AI is used. The law doesn't set a fixed standard.

Where should we start?

Map where AI is used across your organisation. Then check who uses it, what they understand about it, and who reviews its decisions. That shows you where your gaps are.

What are prohibited AI practices?

Prohibited AI practices are uses of AI that the EU AI Act bans outright. There’s no compliance route for them; you simply can’t use them in the EU. The current list covers AI that:

  • Manipulates people using subliminal or deceptive techniques that cause them harm.
  • Exploits vulnerable people, for example targeting someone because of their age, disability or financial situation.
  • Social scoring: rates people on their behaviour or personality and treats them unfairly as a result.
  • Predicts crime based only on someone’s profile or personality traits.
  • Scrapes faces from the internet or CCTV to build facial recognition databases.
  • Reads emotions at work or school, except for medical or safety reasons.
  • Infers sensitive traits from biometric data, such as race, religion, sexual orientation or political views.
  • Identifies people in real time in public spaces for policing, apart from a few narrow exceptions.

‍

Unlock the power of your data & AI

Speak with us to learn how you can embed org-wide data & AI fluency today.