> For the complete documentation index, see [llms.txt](https://experience-digital-transform.gitbook.io/experience-digital-transformation/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://experience-digital-transform.gitbook.io/experience-digital-transformation/regulations/component-5/ai-in-education-governance.md).

# AI in Education Governance

Explainability, human oversight, and appeal rights for AI-influenced educational decisions, the inferred-data challenge, and the case for interim guidance ahead of comprehensive AI legislation.

AI systems are entering education faster than any previous technology, and they concentrate the sector's hardest regulatory questions. AI-assisted tutoring, automated and semi-automated assessment, learning analytics, early-warning systems for dropout, and generative tools for content production are moving from experimental to operational status. What distinguishes them from earlier EdTech is that they do not merely record educational decisions — they make or substantially influence them. The risks this area mitigates are decisions about a child's educational pathway that cannot be explained or appealed, the amplification of existing inequalities by systems trained on historical data, and the accumulation of inferred knowledge about learners that no one consented to create.

AI systems that make or substantially influence decisions about a student's educational pathway — recommending content, flagging learning difficulties, assessing competency — require the same explainability, human oversight, and appeal rights that apply to AI systems in other high-stakes government contexts. This is not a hypothetical standard: under the EU AI Act, AI systems used in education and vocational training for admission, evaluation, monitoring, and proctoring are classified as high-risk, with corresponding obligations for risk management, data governance, transparency, human oversight, and accuracy. Jurisdictions designing their own frameworks can treat this classification as an internationally tested reference point rather than starting from a blank page.

AI-driven learning platforms also generate a particularly acute data protection challenge: they produce **inferred data** — assessments of a student's learning style, emotional state, attention level, or predicted performance — that the student never provided and may not be aware of. The regulatory framework needs to address inferred data explicitly, including whether it may be stored in the learner's record at all, and whether it can be shared with schools, parents, employers, or other parties. This is the point where AI governance and the [student data protection framework](/experience-digital-transformation/regulations/component-5/student-data-protection.md) meet.

AI systems trained on historical educational data may perpetuate and amplify historical inequalities, particularly those related to gender, socioeconomic background, geography, and ethnicity. The regulatory framework should therefore require algorithmic impact assessments for AI systems used in educational decision-making — as a procurement and deployment condition, not an aspiration. The practical instruments are procurement conditions (no AI system influencing educational pathways without an impact assessment and a designated human decision-maker), transparency obligations towards students and parents, and evaluation requirements that link deployment at scale to demonstrated educational value — the point where AI governance connects directly to the [outcome-evidence question](/experience-digital-transformation/regulations/component-5/the-charter-for-public-digital-learning-platforms-and-country-scenarios.md) discussed in the Charter and country-scenarios chapter. One practical corollary deserves emphasis: ministries need not — and should not — wait for comprehensive AI legislation. Interim guidance on classroom use of generative AI, assessment integrity, and vendor transparency can be issued under existing administrative powers, and its absence leaves schools, teachers, and students exposed to inconsistent improvised practice.

{% hint style="info" %}
**EXAMPLES IN PRACTICE — AI in education instruments**

The EU AI Act classifies AI systems used in education and vocational training for admission, evaluation, monitoring, and proctoring as high-risk (Annex III), with obligations phased in through 2025–2027 — a binding template that other jurisdictions can adapt. UNESCO's Guidance for Generative AI in Education and Research (2023), the first global guidance in this space, recommends validating tools for pedagogical appropriateness and setting age thresholds for classroom use. Together they give a regulator both a legal template and a normative checklist.
{% endhint %}

For the fuller comparative treatment of AI governance — the three foundational questions every government must answer, the four global governance approaches, and the Council of Europe's treaty anchor — see [Component 3's AI Governance profiles](/experience-digital-transformation/regulations/component-3-comparative-analysis-of-regulations/ai-governance-regulations-global-landscape-and-best-practice.md), including the [Regulatory Reference and Education page](/experience-digital-transformation/regulations/component-3-comparative-analysis-of-regulations/ai-governance-regulations-regulatory-reference-and-the-education-sector.md) and the [Category 5 AI governance](/experience-digital-transformation/regulations/component-1-the-landscape-of-digital-transformation-regulations/track-a-cross-cutting-guardrails.md#ai-governance-regulation) landscape entry in Component 1.

This completes the five regulatory-area profiles. [The Charter for Public Digital Learning Platforms and Country Scenarios](/experience-digital-transformation/regulations/component-5/the-charter-for-public-digital-learning-platforms-and-country-scenarios.md) maps these five areas — together with connectivity — against a minimum viable regulatory foundation, a set of normative principles, and three country entry-point scenarios.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://experience-digital-transform.gitbook.io/experience-digital-transformation/regulations/component-5/ai-in-education-governance.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.