Current to 10 August 2026 · Not legal advice

Make the EU AI Act
workable.

A plain-language starting point for Irish organisations that need to understand their role, inventory AI uses and translate agreed obligations into delivery practice.

Start with your role

Provider, deployer—or both?

An organisation’s obligations depend on what it does with an AI system. A provider develops a system or places it on the market under its name. A deployer uses an AI system under its authority. Importers, distributors and product manufacturers may have additional roles.

Begin with an AI inventory recording the system, purpose, affected people, supplier, data, decisions, owner and operating context. Do not classify risk from a product label alone.

Practical orientation

Questions every organisation should be able to answer

01

What is the intended purpose?

Define the users, affected people, decisions and prohibited uses.

02

What role do we hold?

Establish whether the organisation is a provider, deployer or another operator.

03

What could the impact be?

Consider safety, fundamental rights, transparency and sector-specific consequences.

04

Who is accountable?

Name business, technical and risk owners, including the authority to stop use.

05

What evidence exists?

Record data, testing, limitations, human oversight, monitoring and supplier information.

06

What happens when it changes?

Plan for incidents, model updates, supplier changes, reassessment and retirement.

Application timeline

A phased framework, not one deadline

The Act is now broadly applicable, but important exceptions and later dates remain. Each organisation still needs to determine its role, use cases and applicable obligations.

1 August 2024

The EU AI Act entered into force.

2 February 2025

The first prohibited-practice rules and the Act’s original AI-literacy provisions began to apply.

2 August 2025

Governance provisions and obligations for general-purpose AI models began to apply.

27 July 2026

The AI Omnibus entered into force, simplifying some requirements and extending the application dates for specified high-risk systems.

2 August 2026

The Act became broadly applicable. Article 50 transparency obligations began to apply, and the AI Office and national authorities began enforcement, subject to specific exceptions and transitional arrangements.

2 December 2027

Requirements for high-risk systems in the Annex III areas identified by the Commission are scheduled to apply.

2 August 2028

Requirements for high-risk systems embedded in regulated products under Annex I are scheduled to apply.

Use primary sources

Regulatory content changes. Confirm the current position using the European Commission’s AI Act portal and the official legislation on EUR-Lex.

Where TechEvolveAI fits

From agreed requirement to operating practice

TechEvolveAI supports delivery and governance work: AI inventories, ownership models, intake processes, evidence planning, review gates, human oversight and production-readiness activities.

TechEvolveAI does not provide legal advice or certify compliance. Legal interpretation and formal assurance should be provided by appropriately qualified specialists.

Open to enterprise engagements

Need a practical AI policy and staff-training baseline?

Explore the fixed-price AI Act readiness check and policy-and-training programme for Irish organisations using AI at work.

Explore AI Act support