EdTech AI compliance under the EU AI Act

Education technology platforms that use AI to assess student performance, adapt learning pathways, detect engagement or disengagement, or monitor student behaviour are subject to some of the EU AI Act's most protective provisions. Annex III Category 3 covers AI systems intended to be used to determine access to, or admission or assignment to, educational or vocational training institutions, as well as systems used to assess students. Because these decisions affect young people's educational futures, the Act treats them as high-risk with a compliance deadline of December 2, 2027. If your EdTech product is deployed in the EU, understanding these obligations now is essential.

What the EU AI Act requires

High-risk EdTech AI systems must comply with Articles 9 through 17. Article 9 requires a documented risk management system covering the full lifecycle of each AI component. Article 10 requires that training and validation datasets are representative, free of known errors where possible, and assessed for bias, which is especially important when AI systems are trained on student data that may reflect historical educational inequalities. Article 11 mandates technical documentation maintained and updated throughout the system's life. Article 13 requires transparency so that educators can understand and interpret AI outputs rather than simply following automated recommendations. Article 14 requires genuine human oversight, meaning a teacher or administrator must be able to review and override AI-generated assessments or routing decisions. Additionally, Article 50 introduces AI content transparency obligations from August 2, 2026, which may affect EdTech products that use AI to generate learning content or provide AI tutoring, as users must be informed when they are interacting with AI.

What this means for your business

An adaptive learning platform that routes students into different curriculum tracks based on AI assessment must be able to demonstrate to schools and regulators that the routing algorithm does not systematically disadvantage students from particular backgrounds. A proctoring tool that uses AI to flag potential academic dishonesty must have documented false-positive rates and a human review step before any sanctions are applied. Schools and universities procuring EdTech are increasingly including EU AI Act compliance as a tender requirement, making this a commercial necessity as much as a regulatory one.

Steps to get compliant

1. Map your Annex III exposure. Identify which AI features affect student admission, assessment, or progression and confirm high-risk classification. Features that only support teachers without influencing outcomes may fall into lower risk categories.
2. Audit your training data for representativeness. Conduct a formal review of the data used to train assessment or routing models, documenting sources, coverage across student demographics, and known limitations as required by Article 10.
3. Implement and document human oversight. Ensure every AI-generated outcome affecting a student's progression can be reviewed and overridden by an educator. Document this workflow in your instructions for use.
4. Prepare Article 50 disclosures. If your product generates AI content presented to students or uses AI tutors, design clear disclosure mechanisms and update your terms of service ahead of the August 2, 2026 deadline.

Free EU AI Act risk assessment

Not sure where your AI system stands? The free ActComply risk screener classifies your system in under 5 minutes. No sign-up required.