The AI Act is a risk-based rulebook. It does not treat every use of artificial intelligence as equally dangerous. The higher the possible impact on people's safety, rights or access to important services, the stronger the obligations become.

Some practices are prohibited. Other systems have transparency duties, including telling people when they are interacting with AI or helping them recognise AI-generated or manipulated content. General-purpose AI models have their own provider obligations, while many high-risk use cases have a later timetable.

• Prohibited practices: certain uses considered incompatible with EU fundamental rights.
• Transparency: people should be able to recognise certain AI interactions and generated or altered content.
• General-purpose AI: model providers have duties around documentation, evaluation and systemic risk where the rules apply.
• High-risk systems: stricter controls apply to listed uses such as parts of employment, education, critical infrastructure and biometric identification.

2 August 2026 was the main application date for the Act. From then, the AI Office and national authorities could begin enforcing the rules that had reached their application date, including the transparency requirements in Article 50 and obligations for general-purpose AI.

The practical effect is not that every AI tool suddenly became illegal or received a government approval label. It is that providers and deployers now have defined duties, and people have clearer routes to ask who is responsible when a regulated AI use causes a problem.

Ireland chose a distributed implementation model rather than one regulator for every AI question. The AI Office of Ireland is the central coordinating authority and single point of contact for the EU AI Act. It helps coordinate competent authorities, provides access to technical expertise and supports innovation, including regulatory sandboxes.

Existing regulators keep the work closest to the sector involved. The published Irish list includes bodies such as the Data Protection Commission, Coimisiún na Meán, the Central Bank, the CCPC, the Health and Safety Authority and the Health Products Regulatory Authority. A regulator's role depends on the AI system and the law it already enforces.

These dates matter because an AI headline can describe a proposal, a rule that already applies, or a requirement that is still in transition. The European Commission's implementation timeline is the best place to check which version is current.

The Act is European law, but its effects are administered in Ireland. Irish Ministers negotiate parts of the EU framework through the Council, Irish regulators enforce within their remits, and the Oireachtas may debate national implementation, funding or related legislation. During Ireland's 2026 EU Presidency, an International AI Summit in Dublin is also bringing the subject into public view.

When an AI issue reaches the Dáil, the useful question is often not simply whether someone is for or against AI. Ask which rule is being discussed, which authority is responsible, what evidence is being requested and whether the proposal changes rights, oversight or public spending. DáilDex can help you follow those questions back to the official record.