Education & Vocational Training , Annex III Point 3 High-Risk AI
AI used to determine access to education, evaluate learning outcomes, steer learners between institutions or programmes, or monitor prohibited behaviour during tests is high-risk under Annex III Point 3. Emotion recognition in educational institutions is additionally prohibited under Article 5.
Scope, Definitions and Boundary with Prohibited Practices
Annex III Point 3 addresses AI in education and vocational training along four specific axes: admissions and enrolment decisions, evaluation of learning outcomes (including for steering learners between institutions), assessment of appropriate level of education an individual will receive, and detection of prohibited behaviour during tests (proctoring).
This is one of the domains where Annex III sits immediately next to an Article 5 prohibition: emotion recognition in educational institutions is prohibited outright by Article 5(1)(f). Proctoring tools that infer stress, confusion, or deception from facial expressions therefore risk falling on the wrong side of the line. The high-risk/prohibited boundary is determined by whether the AI infers emotional state, not by marketing language.
Example Systems Under Annex III Point 3
Illustrative examples, each Member State's market surveillance authority may refine classification guidance over time.
Example
University admissions scoring algorithm
Point 3(a) high-risk, determines access to education and vocational training.
Example
Automated essay grading
Point 3(b) high-risk, evaluates learning outcomes and can steer subsequent placement.
Example
AI-powered adaptive learning platform that recommends course paths
Point 3(c) high-risk where it assesses the appropriate level of education to be received.
Example
Online exam proctoring (behavioural anomaly detection)
Point 3(d) high-risk, monitors prohibited behaviour. If it uses emotion recognition, Article 5 prohibition applies.
Example
Language-learning chatbot with no evaluation function
Out of scope of Point 3, may still trigger Article 50 transparency obligations.
The Articles That Apply
High-risk classification under Annex III triggers the full Chapter III, Section 2 obligation set. Deployers pick up additional obligations under Chapter III, Section 4.
| Article | Obligation | What It Means in Practice |
|---|---|---|
| Art. 9 | Risk Management System | Particular attention to bias affecting protected groups (age, socioeconomic status, language background). Educational AI is one of the areas where algorithmic bias has the most durable downstream effect. |
| Art. 10 | Data Governance | Training data representativeness must cover the learner population the system will actually serve. Historical admissions data frequently encodes past discrimination. |
| Art. 13 | Transparency | Learners (as affected persons) have rights under Article 26(11) and Article 86, the right to explanation of individual decision-making applies. |
| Art. 14 | Human Oversight | A meaningful human review must exist, not merely a rubber-stamp. For admissions, the human oversight often sits at the appeals stage; the Article 14 requirement is that it is designed in, not bolted on. |
| Art. 26(5) | Logging by Deployers | Deployers must keep logs of the system operation for at least six months, educational institutions are often under-prepared for this. |
| Art. 26(6-7) | Worker Information | Where used in workers' training or evaluation, workers and their representatives must be informed. Intersects with the Employment domain (Point 4). |
| Art. 27 | FRIA | Public-sector educational institutions (and private bodies providing public services) must conduct an FRIA before deployment. |
| Art. 49 | EU Database Registration | Registration in the EU database is mandatory before placement on the market. |
Pitfalls to Avoid on Annex III Point 3
Recurrent patterns seen in pre-enforcement readiness assessments.
Proctoring that infers emotional state
If the system classifies stress, confusion, deception or other emotional states, even as an internal flag, Article 5(1)(f) prohibits it in educational contexts. Behavioural pattern detection that does not infer emotion remains high-risk but permitted.
Private-tutoring platforms assuming they are out of scope
If the AI 'evaluates learning outcomes', Point 3(b) applies regardless of whether the setting is public, private, or consumer-grade.
Missing the learner's right to explanation
Article 86 provides affected persons with a right to a meaningful explanation of individual decision-making by high-risk systems. This applies to admissions and evaluation outcomes.
Confusing content-delivery AI with evaluative AI
Recommending course content is not the same as assessing the appropriate level of education. The latter is in scope; the former is generally not unless it steers admission or placement.
Operationalise Point 3 Compliance
The Full Readiness Bundle gives your legal and compliance teams the 58-Point Compliance Checklist, Annex III Classification Matrix, FRIA template, Annex IV Technical Documentation Checklist, and 7 more documents, all aligned to the Annex III high-risk regime (now applying from 2 December 2027 under the May 2026 Omnibus deal).
one-time · instant download
Get the Full Bundle →Also available: Checklist Pack $149 · White-Label $999/yr
Next Steps
Deadline Planning
15-week compliance work plan for the EU AI Act's phased timeline.
Week-by-Week Plan →