← Blog
Education April 17, 2026 · 13 min read

AI Compliance for Education: FERPA, EU AI Act, and Student Data Requirements

Education is one of the highest-risk sectors under the EU AI Act — AI used in admissions, assessments, and student placement is explicitly classified as high-risk under Annex III. Schools, universities, and EdTech companies must navigate FERPA, COPPA, EU AI Act, and an expanding set of state student privacy laws simultaneously.

Why AI in Education Is High-Stakes for Compliance

Three converging factors make education a compliance priority:

Minors are the primary subjects

Children and young adults have heightened data protection rights under every jurisdiction. COPPA (under-13s), GDPR (special protections for minors), and state laws all impose stricter requirements when AI processes student data.

Consequential decisions with lifelong impact

AI systems in education often make or influence decisions about admissions, grade progression, academic support, and discipline — decisions that affect students' life trajectories. Regulators treat these as high-consequence AI use cases.

Legacy data sharing practices

Education has historically shared data widely with EdTech vendors under broad FERPA exceptions. AI changes this risk calculus: vendors can now infer far more from data than FERPA's drafters anticipated in 1974.

EU AI Act Risk Classification for Educational AI

The EU AI Act Annex III para.3 explicitly lists "AI systems intended to be used for the purpose of determining access to or assigning persons to educational and vocational training institutions, or to evaluate learning outcomes" as high-risk.

AI for student admissions or placement decisions

High-Risk (Annex III)

Why: Access to education and educational benefits — explicitly listed in Annex III para.3(b)

Required: Conformity assessment, technical documentation, human oversight, bias testing

AI that assesses or grades student work

High-Risk (Annex III)

Why: Evaluation of learning outcomes determines educational opportunities

Required: Full high-risk AI requirements including explainability and audit logs

AI that detects student emotional state or engagement

High-Risk (Annex III)

Why: Biometric-adjacent; infers psychological state; affects minors

Required: Strict data minimisation, parental consent, DPA required

Adaptive learning platforms that personalise curriculum

Limited Risk or High-Risk

Why: May be high-risk if it shapes educational trajectory; limited risk if purely presentational

Required: Risk assessment required to determine applicable tier

AI chatbots for student support (mental health)

High-Risk (contested)

Why: Mental health support may qualify as AI in healthcare under Annex III para.5

Required: Consult legal counsel; default to high-risk requirements as precaution

AI-generated content tools for student use (writing assistants)

Limited Risk

Why: No consequential individual decision-making

Required: Transparency: disclose AI use to students and staff

August 2, 2026 deadline

High-risk AI systems must meet full EU AI Act requirements by August 2, 2026. Schools and EdTech companies operating in the EU should begin conformity assessments and technical documentation immediately if they have not already.

FERPA and AI: Five Rules Every Institution Needs to Know

The Family Educational Rights and Privacy Act (FERPA) was written in 1974 — long before AI. The US Department of Education has not issued AI-specific FERPA guidance as of 2026. However, the existing rules apply clearly to most AI use cases in education:

1

Education records are protected

Any record directly related to a student and maintained by an educational institution is an "education record" covered by FERPA. This includes AI-generated assessments, performance predictions, and behavioural flags.

2

Third-party AI vendors need written agreement

Schools can share education records with AI vendors if they qualify as a "school official" with a "legitimate educational interest" — but only under a written agreement that restricts how the vendor uses the data.

3

Training AI on student records without consent is likely a violation

Using student education records to train an AI model — even for improving the educational service — is not clearly covered under the school-official exception. FERPA consent may be required.

4

Parental rights until age 18

For K-12 students, FERPA rights belong to parents. Any AI system that processes student education records must accommodate parental access, correction, and consent rights.

5

State laws may be stricter

SOPIPA (California), New York Education Law §2-d, Colorado HB 20-1468, and similar laws impose additional restrictions on EdTech vendors beyond FERPA. Always check state law.

COPPA and Student Privacy Checklist for AI Systems

COPPA applies to AI systems that collect personal information from children under 13. For schools using AI, the school-consent exception allows teachers and school officials to consent on behalf of parents — but only for services used for educational purposes, not for commercial AI products.

No AI system collects personal data from children under 13 without verifiable parental consent

No AI system uses student data for targeted advertising or commercial profiling

Third-party AI vendors have agreed not to use student data beyond providing the contracted service

Data minimisation: AI systems collect only data necessary for the educational purpose

Parents and students (13+) can access, correct, and delete AI-generated records about them

All AI vendors who access student data are listed in the school's annual FERPA notification

Key US State Student Privacy Laws Affecting AI

SOPIPA (California)

Prohibits EdTech vendors from using student data for targeted advertising, building behavioural profiles for non-educational purposes, or selling student data. Covers operators of school-directed services — including AI tools.

New York Education Law §2-d

Applies to all "third-party contractors" handling student data. Requires Parents' Bill of Rights notice, strict data security requirements, and limits use to the contracted educational purpose. One of the strictest state-level protections.

Colorado AI Act (SB 24-205)

Effective June 30, 2026. If an AI system is used in student placement or educational benefit decisions, Colorado's deployer obligations apply — including impact assessments and student notification rights.

Illinois AIVIA and BIPA

Biometric Information Privacy Act applies to AI systems that analyse facial geometry, voiceprints, or other biometrics — including emotion-detection or engagement-monitoring AI in classrooms.

For EdTech Companies: What You Must Build In

1

Data Processing Agreement template that meets both FERPA school-official exception and GDPR Art.28 requirements

2

Data use limitation: contractual and technical controls preventing use of student data beyond the contracted service

3

Model training disclosure: clear statement in DPA of whether student data is used for model training, and if so, how

4

Parental access portal: mechanism for parents to access, correct, and delete AI-generated records about their child

5

Annual audit rights: allow school clients to audit compliance with data use restrictions

6

EU AI Act conformity documentation for high-risk AI products — required for EU school customers

7

Human oversight mechanism: high-risk AI decisions must be reviewable by a teacher or counsellor before taking effect

Map Your AI Compliance Obligations

ComplianceIQ automatically identifies which regulations apply to your AI systems based on your sector, jurisdiction, and use case — including FERPA, EU AI Act, and COPPA.

Run a Free Risk Assessment