Can Teachers Use AI in Irish Schools? What's Allowed and What's Banned?

The question being asked in staffrooms is often the same one a principal wonders about: are we actually allowed to use this? The answer is yes. The Act has viewpoints, but banning teachers from using AI isn't one of them. What it does do is ban specific uses of AI in educational settings and require that AI use is governed rather than improvised.

The Act has viewpoints, but banning teachers from using AI isn't one of them

What does the EU AI Act actually prohibit in schools?

Article 5 of the EU AI Act, which applies to Irish schools as deployers, lists practices that are banned outright. These prohibitions have been in force since 2 February 2025. Four apply directly in schools.

Emotion recognition

AI that reads students' or teachers' emotions by analysing physical signals is banned in schools. This covers any system that watches faces, analyses voices, or uses sensors to draw conclusions about how someone is feeling. A tool that observes students' faces during a lesson and decides whether they look engaged, bored, or distracted is exactly what this targets.[1] AI that reads the tone of something a student has written is not in the same category; the ban is about physical signals, not written ones. The European Commission published guidance clarifying this distinction on 4 February 2025.[2]

Manipulation of minors

AI designed to exploit children's psychological vulnerabilities to change their behaviour is banned. This is not about tools that encourage learning or reward effort. It's about systems specifically built to override a child's own choices using techniques they may not even notice. A learning platform that detects when a student is about to disengage and responds with prompts like "three classmates are ahead of you" or "your streak ends tonight" is using this tactic.[3]

Social scoring

AI that builds up a picture of how a student behaves over time and uses it to affect decisions in a completely different context is prohibited.[4] The concern is not behaviour monitoring itself; it's when monitoring generates an automated classification that follows the student beyond the school. If a management system automatically produces a conduct profile from years of incident records, and that profile travels with a student's file to their post-primary school, the child arrives already categorised by an algorithm. That is what this prohibition is designed to prevent.

Biometric identification

AI that identifies people by scanning their physical features in real time is generally prohibited.[5] In a school context this means camera systems that automatically recognise students' faces, whether they are marketed as attendance tools, security systems, or safeguarding solutions. Some of these systems are actively sold to schools. Regulators have already ruled against schools that trialled them, finding that children cannot meaningfully consent to being identified by a camera whenever they walk through a door.[6]

Most Irish schools are not currently using any of these tools and so do not fall within any of the prohibitions. ChatGPT, Microsoft Copilot, and Google Gemini are not emotion recognition systems. The Article 5 prohibitions are aimed at a different category of AI entirely.

Can teachers use ChatGPT and similar types of tools?

Yes. Using a generative AI tool to draft a lesson plan, create quiz questions, produce specific resources for a mixed-ability class, or reduce administrative load is not prohibited under the EU AI Act. The Department of Education has indicated that there's no obligation on schools to use AI;[7] that's a separate discussion from whether teachers are actually permitted to use these tools.

The Act sets limits on how AI can be used and requires that use to be overseen. Most teachers' use of generative AI sits well clear of those bounds.

What does the Act actually require of Irish schools?

Entering personal student data into any AI tool creates a data protection obligation. The tool is not the question; the data is.

Twinkl's AI Report Writer illustrates this precisely. The Report Writer generates pupil summaries from information the teacher provides, including the child's name, subject performance, and targets.[8] That means personal student data is entering a third-party AI system. For that to be compliant, the school needs a data processing arrangement with Twinkl that specifically covers its AI tools. Using Twinkl for lesson resources raises no significant concern; that use involves no student data. The Report Writer does.

The tool is not the question; the data is

This applies to every AI tool used with real student information. The obligation follows the data, not what the tool does.

Beyond data protection, the Act has a second category of obligations.

The Act requires schools, as deployers of AI systems, to ensure a sufficient level of AI literacy among staff. In practice that means staff knowing where AI genuinely helps, where it falls short, and what to keep an eye on when they use it. The Act also asks that prohibited practices are understood and avoided, and that there is oversight of AI operating under the school's authority. Updating the Acceptable Use Policy (AUP) to address AI, and running a structured staff literacy session, are the steps that turn a permission question into a resolved one.

Read more: Do Irish schools need an AI policy? What the AUP does and doesn't cover →

What about the tools staff are actually using?

The most common AI tools in Irish schools fall into three broad categories for governance purposes.

School-deployed tools with enterprise agreements

Microsoft Copilot and Google Gemini, accessed through school accounts under Microsoft 365 Education and Google Workspace for Education respectively, operate within data processing agreements the school already holds. For teacher-facing use, the risk profile is lower: those agreements govern what happens to data entered into the tool.

Teacher-managed tools with available data protection arrangements

Canva for Education is free for verified teachers[9] and provides a GDPR-compliant Data Processing Addendum for EU users, with Canva acting as data processor on the school's behalf.[10] It is verified by the individual teacher using a school email rather than deployed at school level. The data protections are meaningfully better than a personal consumer account, but the school may have no formal record of staff using it.

Twinkl is available to Irish primary schools as a resource and planning platform.[11] For general lesson resources, the governance position is similar to Canva: teacher or school-managed rather than enterprise-deployed. The AI Report Writer changes the picture. Each school should verify that Twinkl's data processing arrangements cover its AI features before staff use the Report Writer with pupil data.[12]

Personal accounts

ChatGPT accessed through a personal account has no data processing agreement in place. Staff using personal accounts are operating outside the school's data governance arrangements entirely. That is reason to address it in the AUP, not to prohibit the tool.

Read more: Your teachers are already using AI. The question is whether you know about it →

There is one edge case worth knowing about. The EU AI Act distinguishes between schools using a general AI tool and schools deploying that tool to drive significant decisions about individual students. Automated grading, AI-filtered admissions, or AI-weighted assessments would cross that line. A school making that choice takes on obligations closer to those of the companies that build these tools: conformity assessments, technical documentation, and a substantially heavier compliance burden overall. Everyday teacher use such as lesson planning, resource generation or administrative work sit clear of the line. This only becomes relevant if your school is considering using AI to make or weight decisions that directly affect students' progress. If that conversation is happening, it warrants a separate assessment before anything goes live.

In a nutshell

Teachers can use AI. There never was a rule against it

Teachers can use AI. There never was a rule against it. What most schools still need is the framework that allows an overview and governance: a clear AUP, a staff session on what is and is not permitted, and a working picture of what tools are already in use.

Read more: Three steps to get your school started on AI governance →

The permission debate is pretty much settled. The governance question is where most schools are still catching up.