AI and Your School: what every international school should know

Artificial intelligence is already in your school. It may be in your admissions process, your assessment tools, your plagiarism detection software, or simply in the devices your teachers and students use every day. Most school leaders know this. Fewer know that the legal framework governing how schools use these tools has changed substantially, and that understanding your school's position under the EU AI Act is now a governance question, not a future agenda item.

This article is not a legal briefing. It is a practical guide for Heads, principals, and governors who need to understand what is coming, why it matters for international schools specifically, and what to do now.

What the EU AI Act actually says about schools

The EU Artificial Intelligence Act, Regulation (EU) 2024/1689, entered into force in August 2024 and is being implemented in phases. The most significant phase for schools takes effect on 2 August 2026, when the comprehensive compliance framework for high-risk AI systems becomes enforceable.

Education is explicitly named in the Act. Under Annex III, AI systems used in schools to determine admissions, evaluate learning outcomes, assess appropriate levels of education, or monitor student behaviour during tests are classified as high-risk. This is not a broad interpretation; it is the text of the legislation.

High-risk classification carries concrete obligations. Schools deploying such systems must maintain technical documentation, implement human oversight mechanisms, keep automated logs of AI decisions, and be able to demonstrate compliance to regulators on request.

A word of caution. The European Commission missed its own February 2026 deadline to publish implementation guidance for Article 6, and the regulatory picture has shifted further since then. The Digital Omnibus, a legislative package that proposes extending the high-risk compliance deadline from August 2026 to December 2027 for Annex III systems, passed the European Parliament's IMCO and LIBE committees in March 2026 by 101 votes to 9, and is on track for adoption before the original August deadline arrives. By the time you read this, the extension will very likely already be law. This does not change the governance calculus for schools. The obligations themselves (documentation, human oversight, risk classification, staff training) have not changed; only the enforcement date has moved. A board that has done nothing on AI governance because a deadline shifted is a board that has confused regulatory enforcement with governance responsibility. Waiting for legal certainty before acting is not a posture many will find satisfactory.

Does this apply to your school?

The Act has extraterritorial reach. Under Article 2, it applies wherever the output of an AI system is used in the EU, but also to any provider or deployer established or located in the EU, regardless of where their school physically sits. If your school is part of a group with EU operations, or if student data and AI-generated decisions flow back through EU-based systems, you are likely in scope. If your school operates entirely outside the EU with no EU-based infrastructure, the position is less clear. But given that many international school families are EU nationals and that data flows are rarely simple, the safest governance assumption is that you should understand your exposure before concluding you are exempt. This is a question for your legal adviser.

This will be unfamiliar ground for many international school boards. Safeguarding legislation, child protection frameworks, and data protection requirements have historically been the compliance landscape governors carry. AI governance is now joining that list. The question is not whether your school uses AI. The question is whether you can evidence that you are governing it responsibly.

What inspectors and governors will start asking

BSO inspection requires British international schools to demonstrate effective governance across all aspects of school operation. Many of those same schools also hold COBIS membership and CIS accreditation, each of which carries its own standards expectations around governance. As AI becomes embedded in teaching, assessment, and administration, the expectation that schools can articulate their AI governance position, in writing and with evidence, will follow naturally across all three frameworks.

Governors have a specific accountability here. A board that cannot answer basic questions about which AI systems the school uses, how student data is protected within those systems, and what oversight mechanisms are in place is a board that is not fulfilling its governance responsibilities. That is not a future concern. It is a present one.

Five things every international school should have documented before August 2026

1. An AI inventory. A simple register of every AI tool in use across the school: administrative, pedagogical, and pastoral. Who uses it, what data it processes, and whether it falls into the high-risk categories defined by Annex III. Most schools do not have this. It takes a structured half-day to produce a first version.

2. A risk classification. For each tool in the inventory, a documented assessment of its risk level under the Act. This does not require a lawyer. It requires someone who understands the framework and the school's specific context. Where a tool is used for admissions screening, assessment, or student monitoring, the default assumption should be high-risk until assessed otherwise.

3. An AI policy that staff, students, and parents have seen. Not a document filed in a shared drive. A policy that has been communicated, discussed in staff meetings, included in the student handbook, and shared with parents. The communication trail matters as much as the policy itself.

4. Staff training records. Evidence that every teacher has received basic training on responsible AI use, academic integrity in an AI context, and safeguarding implications. Completion records must be retained.

5. Governor oversight framework. A standing agenda item for AI governance at board level, with termly reporting from the headteacher on AI use, incidents, and policy compliance. Governors who have never discussed AI at a board meeting are governors who cannot evidence oversight.

A note on international schools specifically

COBIS member schools operate across jurisdictions with varying regulatory environments. The EU AI Act's reach is broad but its application to schools outside the EU depends on specific circumstances: whether EU residents are among your students, whether your school is part of a group with EU operations, and how your data flows are structured. These are questions worth exploring with your legal adviser rather than assuming you are either fully in scope or entirely exempt.

What is not jurisdiction-dependent is the governance principle. Whether or not your specific school is legally required to comply by August 2026, a board that has done nothing on AI governance is a board that is behind the curve of what COBIS, CIS, and IB will increasingly expect from their accredited schools.

Where to start

The five steps above are achievable without specialist technical knowledge. They require leadership time, structured thinking, and the willingness to have honest conversations about what AI is actually doing in your school.

Education Options publishes AI Policy Packs for schools that need a ready-made framework. Schools that need bespoke support to work through the inventory, risk classification, or policy drafting can book a complimentary consultation.