Position checked 4 November 2026
We’d start by listing what the AI actually does. Does it draft an email for staff, answer questions from customers or rank job applicants? The answer affects what you need to check, even if the tools come from the same supplier.
Where the rules stand in Ireland
Irish position as of 4 November 2026. The EU AI Act applies directly in Ireland according to the EU timetable below. The Regulation of Artificial Intelligence Act 2026, signed on 21 July, established the AI Office of Ireland as the central coordinator and single point of contact. Oversight is shared across designated authorities, with the relevant authority depending on the use or sector. Irish Act and coordinating role · Irish competent authorities
The EU AI Act entered into force in August 2024 and has applied in stages. AI literacy duties and the original bans on certain practices have applied since 2 February 2025. Most provisions, including the general transparency rules, have applied since 2 August 2026.
Under the amended timetable, the high-risk rules for Annex III systems, which include certain recruitment tools, apply from 2 December 2027. The rules for qualifying high-risk AI systems covered by EU product legislation apply from 2 August 2028. EU amending regulation and timetable · AI Office of Ireland
1. If your staff use AI for office work
If you use a supplier’s AI tool in your business, you’re generally a deployer. If you develop a system or have one developed and place it on the market or put it into service under your own name, you may be a provider. Check your role for each use, especially if you’re building a tool in-house or commissioning one under your brand. AI Act definitions
Article 4 requires providers and deployers to take measures to support AI literacy among staff and others using AI on their behalf. For a team using AI to draft or summarise documents, a practical starting point is to explain which tools are approved, what information can go into them, who checks the output and where to raise a problem. Tailor that guidance to the people and the work.
Check proposed uses against Article 5 too. For example, it generally prohibits using AI to infer emotions in the workplace, with exceptions for medical or safety reasons. AI literacy rules · Prohibited practices
2. If you run a customer-facing chatbot
Article 50 requires providers of AI systems designed to interact directly with people to make that interaction clear by the first exchange, unless it’s already obvious in context. If you use a supplier’s chatbot, ask how the notice is handled and test what customers actually see on your site. Commission guidance on Article 50
The 2 December 2026 transition applies to machine-readable output marking by providers of certain generative systems placed on the market before 2 August 2026. Chatbot notices follow the August timetable.
Separate disclosure rules apply when businesses use AI to create deepfakes or publish AI-generated or manipulated text to inform the public on matters of public interest. For that text, an exception applies where there has been human review or editorial control and a person or organisation holds editorial responsibility. The disclosure rule turns on how the content is used. Article 50 · Article 111
3. If you use AI to screen CVs
AI that ranks, filters or evaluates applicants is generally high-risk. A narrow exception can apply to certain limited tasks that pose no significant risk to people’s health, safety or fundamental rights. An Annex III system that profiles people is always high-risk.
Ask your supplier for the tool’s intended purpose, its classification and how it plans to meet the applicable requirements. If you already use the tool, check how the transition rules apply to that system and any planned changes. Annex III · Article 6 · Article 111
GDPR applies separately. Before putting applicant, customer or staff data into an AI tool, check your lawful basis, how the supplier handles the data, what you tell people and whether a data protection impact assessment is required. Do that now, even where the AI Act’s high-risk rules apply later. Irish Data Protection Commission guidance · DPIA guidance
Where to start this week
Make a list of your AI tools, what they do, who owns each one and what data goes into it. Use that list to check staff guidance, test customer-facing disclosures and identify any tools that influence decisions about people. For those tools, get the supplier’s classification and intended purpose in writing.
Keep a record you can use when something changes
For each use, record the tool and supplier, who uses it, the intended purpose, what data it receives, whether it affects anyone’s opportunities or rights, and who can override or stop it. Add the supplier’s classification and any notices or instructions. Revisit that record when a tool moves from drafting to making recommendations, or from internal use to interacting with customers.
Professional services: A document assistant drafts a summary. Keep an approved source list and ask a colleague to check citations before sharing it. Record which staff need guidance on confidential material.
Retail or hospitality: Walk through the chatbot as a customer. Check how it identifies itself, what it does when it cannot answer, and how someone reaches a person. Test the actual first interaction after configuration changes.
Recruitment: Distinguish booking interviews from evaluating applicants. If the tool ranks or filters people, ask the supplier to explain its intended purpose and classification before use. Confirm how staff can review decisions and handle applicant data.
Use the dates and examples above to schedule reviews of the AI you use. The phased deadlines are a reason to review the inventory regularly. Amended Article 113 · Article 50 transparency duties · AI Office of Ireland
Start with one AI tool
Pick one tool and write down what goes into it, who checks the results and who could be affected. We can help you work through how the tool is used and what you need to ask the supplier.