Your Exam Rights Under the EU AI Act: Proctoring, Emotion Recognition and the Right to an Explanation

Your Exam Rights Under the EU AI Act: Proctoring, Emotion Recognition and the Right to an Explanation


Picture the exam: you sit at your desk, a camera watches you, and the software decides whether your eye movement, your posture or your face looks "suspicious". You never see the score. You never meet the algorithm. Then a flag lands on your file.

Most students in that position assume they have no rights — that the software is a fact of university life and the only question is whether they cheated. That is wrong. The EU AI Act gives students specific, enforceable rights in exactly this situation. Some of them apply today. Some apply from December 2027. And one of the most important ones — the ban on emotion recognition in education — has applied since 2 February 2025.

This is the rights angle nobody else covers for students: what the law says, what your university owes you, and what to do when an AI system decides unfairly about you.

The ban that already applies: emotion recognition in education

Emotion recognition in education is banned. Not restricted, not high-risk — banned outright. Article 5(1)(f) of the EU AI Act:

"the placing on the market, the putting into service for this specific purpose, or the use of AI systems to infer emotions of a natural person in the areas of workplace and education institutions, except where the use of the AI system is intended to be put in place or into the market for medical or safety reasons"

Applied since 2 February 2025. Breach = the highest fine tier, Article 99(3): up to €35 million or 7% of global annual turnover.

What counts: an emotion recognition system works from biometric data (Article 3(39)) — it identifies or infers emotions or intentions on the basis of biometric data. Sentiment analysis of written text falls outside the ban (no biometric data involved). The exceptions are narrow — medical or safety only. The classic lawful example is monitoring driver or pilot fatigue. A "productivity dashboard" or an "engagement score" is not a safety reason. The exception must be the system's intended purpose, not a side benefit.

The Commission's guidelines on prohibited practices (C(2025) 5052, 29 July 2025) give education examples — all PROHIBITED: inferring students' interest and attention; emotion recognition during admission tests; adding emotion detection, such as arousal or anxiousness, to online exam monitoring.

What this means for you: if your exam software reads your face or voice to infer stress, attention, arousal or "suspicious" emotions, that feature is banned outright — it is not a high-risk system with safeguards, it is a prohibited practice. Proctoring that only records video or flags behaviour (without inferring emotions) is a different use — see the next section.

The high-risk box: proctoring and grading

Annex III point 3 lists four categories of AI in education and vocational training, at all levels, as high-risk:

  • (a) AI systems intended to be used to determine access or admission, or to assign natural persons to educational or vocational training institutions;
  • (b) AI systems intended to be used to evaluate learning outcomes, including when those outcomes are used to steer the learning process;
  • (c) AI systems intended to be used to assess the appropriate level of education that an individual will receive or will be able to access;
  • (d) AI systems intended to be used to monitor and detect prohibited behaviour of students during tests.

Automated grading falls under (b); remote/AI proctoring falls under (d). A tutoring chatbot that only answers questions is NOT high-risk (it carries transparency duties under Article 50 instead).

Nuance — the Article 6(3) exit: a system in Annex III point 3 is NOT high-risk if it performs only a narrow procedural task, only improves the result of a previously completed human activity, only detects decision-making patterns or deviations without replacing or influencing human assessment, or only performs a preparatory task — and never where it profiles natural persons. The provider must document that assessment before placing it on the market (Article 6(4)). Practical line: a tool that only records the exam is likely not high-risk; a tool that flags behaviour, scores attention or grades is.

What the university owes you (Article 26) — the university is the deployer, the edtech company is the provider:

  • Human oversight (Article 26(2)): assign oversight to natural persons with the necessary competence, training, authority and support. The provider-side mirror is Article 14: overseers must be able to monitor, intervene, override or stop the system, and be aware of automation bias.
  • Check the input data (Article 26(4)): the data must be relevant and sufficiently representative for the intended purpose.
  • Monitor and suspend (Article 26(5)): if there is reason to consider the use may present a risk, inform the provider and the relevant market surveillance authority without undue delay and suspend use.
  • Keep the logs (Article 26(6)): keep the automatically generated logs, to the extent under the deployer's control, for a period appropriate to the intended purpose and at least six months, unless other Union or national law provides otherwise.
  • Tell the student (Article 26(11)): "deployers of high-risk AI systems referred to in Annex III that make decisions or assist in making decisions related to natural persons shall inform the natural persons that they are subject to the use of the high-risk AI system".

Fundamental Rights Impact Assessment (Article 27): before the first use of an Annex III high-risk system, bodies governed by public law (this covers most public universities) — and private entities providing public services, which Recital 96 links to tasks in the public interest such as education — must carry out a Fundamental Rights Impact Assessment. It must describe the processes, the timing and frequency, the categories of people affected, the specific risks of harm, the human oversight measures, and the measures to be taken if risks materialise — including internal governance and complaint mechanisms. After the assessment, the deployer notifies the market surveillance authority of the results. Where an obligation is already met through a DPIA under GDPR Article 35, the deployer may cross-reference the relevant DPIA sections — but a DPIA does not replace the FRIA.

The honest date: Chapter III, Sections 1, 2 and 3 (the high-risk requirements and Article 26 deployer duties) apply from 2 December 2027 for Annex III stand-alone systems (Article 113(3)(c) as replaced by Regulation (EU) 2026/1744), and 2 August 2028 for high-risk AI embedded in Annex I regulated products. The duties are law now, but they only bite from December 2027 — until then, GDPR and national exam rules are the live protection.

The two rights that let you push back

Article 86 — the right to an explanation

Exact wording (Article 86(1)) — quote verbatim:

"Any affected person subject to a decision which is taken by the deployer on the basis of the output from a high-risk AI system listed in Annex III, with the exception of systems listed under point 2 thereof, and which produces legal effects or similarly significantly affects that person in a way that they consider to have an adverse impact on their health, safety or fundamental rights shall have the right to obtain from the deployer clear and meaningful explanations of the role of the AI system in the decision-making procedure and the main elements of the decision taken."

Conditions (all must be met): (1) a decision taken by the deployer (the university, not the software vendor); (2) based on the output of a high-risk AI system listed in Annex III (point 2, critical infrastructure, excluded); (3) the decision produces legal effects or similarly significantly affects the person in a way they consider adverse to their health, safety or fundamental rights. Recital 171 adds the decision should be based mainly on the system's output, and the explanation should enable the person to exercise their rights.

Limits: Article 86(2) — the right does not apply where exceptions or restrictions follow from Union or national law in compliance with Union law. Article 86(3) — it applies only to the extent the right is not otherwise provided for under Union law (so it fills gaps rather than duplicating GDPR rights). It is a right to an explanation, not a right to reverse the decision.

The date — be honest about it: the general timetable makes Article 86 applicable from 2 August 2026 (Article 113). It sits in Chapter IX, Section 4 (Remedies), and the Digital Omnibus (Regulation (EU) 2026/1744) neither amended Article 86 nor expressly deferred it. But the Omnibus postponed Article 6(2) and most of the rules by which Annex III systems are legally classified as high-risk until 2 December 2027. So the right's trigger — a decision "based on the output of a high-risk AI system listed in Annex III" — depends on a classification regime that only fully bites in December 2027. No final Commission guidance and no CJEU ruling had settled the relationship as of late 2026. Practical advice: the right is on the books since 2 August 2026 and you can invoke it now — but be honest that its practical trigger is tied to the high-risk regime that fully applies from 2 December 2027, and the timing question is not yet settled.

GDPR Article 22 — the older, live-today protection

Exact wording (Article 22(1), Regulation (EU) 2016/679) — quote verbatim:

"The data subject shall have the right not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects concerning him or her or similarly significantly affects him or her."

  • Three exceptions (Article 22(2)): (a) necessary for entering into or performance of a contract; (b) authorised by Union or Member State law which lays down suitable safeguards; (c) based on the data subject's explicit consent.
  • Safeguards (Article 22(3)) for the contract and consent exceptions: at minimum the right to obtain human intervention, to express their point of view, and to contest the decision. The reviewing human must have the authority, competence and practical ability to change the outcome.
  • Key limit — "solely": Article 22 covers decisions based solely on automated processing. If a genuine human reviews the output, weighs the factors and makes an independent call, Article 22 does not apply. A rubber-stamp "approve" click does not count as meaningful human involvement. CJEU, SCHUFA (C-634/21, 7 December 2023): automated scoring can itself constitute the relevant decision where a third party draws strongly on the score.
  • Related GDPR rights: Article 15(1)(h) — right to meaningful information about the logic involved in automated decision-making; Articles 13(2)(f) and 14(2)(g) — controllers must inform data subjects about the existence of automated decision-making, the logic, and its significance and envisaged consequences. Article 35(3)(a) — a DPIA is mandatory for systematic and extensive evaluation of personal aspects based on automated processing on which decisions with legal effects are based.

How Article 22 and Article 86 interact: if the decision is purely automated AND made by a high-risk AI system, both apply (Article 22 gives the right to human intervention and contestation; Article 86 gives the right to an explanation). If a human genuinely reviewed it, only Article 86 applies. If the AI is not high-risk and a human reviewed it, neither applies — you fall back on university exam-appeal rules and GDPR transparency and access rights.

The practical playbook

Step 1 — inside the university (first, and fastest):

  • Ask for the Article 86 explanation in writing from the deployer — the university or exam office: what the AI system did, what it scored or flagged, and how that fed into the decision. Also ask which tool was used and whether the decision was solely automated (which triggers GDPR Article 22).
  • Use the university's exam-appeal route: the Examination Board or examination committee decides fraud and grade disputes at most EU universities. The course manual or study guide sets the procedure and the deadline — appeal deadlines are usually short (days to a few weeks), so act immediately.
  • Real university practice: Utrecht University — if teachers suspect GenAI use contrary to the course manual, they may ask the student to explain their work orally; if fraud or plagiarism is possible, the teacher must report to the examination committee, which investigates and may impose sanctions. Wageningen University (WUR), assessment policy addendum September 2026: "Lecturers must not use AI to assess student work... You may use it to provide feedback, but AI is too much of a black box to be used for final assessments or awarding marks. The final assessment and the awarding of marks remain the responsibility of the lecturer."
  • Student representation: the student council or participation council and the student ombudsman or dean of students can support an appeal.

Step 2 — evidence to keep (from the moment you are flagged):

  • The exam notice and instructions and the proctoring consent or information letter (what you were told, and when);
  • The flag itself: screenshots, the exact time, what you were doing, the proctoring report if shared;
  • Your work: drafts, version history, notes, sources, prompt logs;
  • All correspondence with the instructor and exam office, with dates;
  • Ask for: the proctoring logs (kept at least six months under Article 26(6) once the regime applies), the DPIA summary, and whether a FRIA was done. Under GDPR you can also file a subject-access request (Article 15) for your personal data, including the logic of any automated decision (Article 15(1)(h)).


Step 3 — outside the university (if the internal route fails or the breach is systemic):

  • AI Act, Article 85 — complaint to the market surveillance authority: "any natural or legal person having grounds to consider that there has been an infringement of the provisions of this Regulation may submit complaints to the relevant market surveillance authority." Each Member State designates its market surveillance authority or authorities (designation was due by 2 August 2025); look up your country's authority — the Commission's AI Act Explorer and national AI Office pages list them.
  • GDPR, Article 77 — complaint to the national data protection authority (for example the Dutch Autoriteit Persoonsgegevens, the German state DPAs, the French CNIL): for the personal-data side — unlawful processing, no legal basis, excessive monitoring, no DPIA. This route is live today and does not depend on the AI Act's high-risk dates.
  • Courts: the leading proctoring precedent is the Amsterdam District Court, 12 June 2020 (UvA vs student councils): UvA could continue using Proctorio during online exams without student consent — the university's public task (education, exams, degrees) is laid down by law, so GDPR allows processing necessary for that task; no live monitoring (invigilators only saw data after the program detected significant abnormal behaviour); data encrypted and destroyed after 30 days. Students' argument that consent was required failed. Honest framing: courts have so far sided with universities on proctoring during the pandemic; the AI Act's new rights (Article 86 explanation, Article 85 complaint) are the new leverage, but no AI Act proctoring case has been decided yet. The Dutch Education Minister (Dijkgraaf, February 2022) declined to intervene nationally: "It is the responsibility of higher education institutions to ensure that privacy is in good order."

What NOT to rely on: consent. Universities generally do not need student consent for proctoring (public-task legal basis under GDPR Article 6(1)(e)); the argument is about necessity, proportionality, transparency and human review — not consent.

2026 timeline — what applies when

DateWhat
2 Feb 2025Emotion recognition in education banned (Art. 5(1)(f)); AI literacy duty (Art. 4)
2 Aug 2025GPAI obligations; governance; penalties
27 Jul 2026Digital Omnibus Regulation (EU) 2026/1744 in force
2 Aug 2026General application + Art. 50 transparency; Art. 86 right to explanation on the books
2 Dec 2026Two new Art. 5 prohibitions (non-consensual intimate imagery; AI-generated CSAM) + Art. 50(2) machine-readable marking deadline
2 Dec 2027Annex III high-risk obligations, including Art. 26 deployer duties and Art. 27 FRIA
2 Aug 2028High-risk AI embedded in Annex I regulated products
2 Aug 2030Public-authority systems (Art. 111)

Enforcement as of 9 November 2026: no public AI Act fine or enforcement decision on proctoring or emotion recognition was found. Market surveillance authorities were designated by Member States (due 2 August 2025) and the AI Office complaints tool is live. The UvA 2020 court case remains the leading proctoring precedent; the AI Act rights (Articles 85 and 86) have not yet been tested in court for exams.

Universities moving on their own: Utrecht University has banned using AI detection software to scan students' work; Leiden also does not allow it (Times Higher Education, 15 October 2026). UNL (Universities of the Netherlands) warned detection tools "are not always accurate in identifying misconduct". These are institutional choices, not AI Act duties — but they show the direction of travel.

Frequently asked questions

Can my university use software that reads my emotions during an exam?
No. Emotion recognition in education has been banned since 2 February 2025 (Article 5(1)(f)), except for medical or safety purposes. A system that infers stress, attention or arousal from your face or voice during an exam is a prohibited practice.

Does my university have to tell me it is using AI proctoring?
Yes — once the high-risk regime applies, Article 26(11) requires deployers to inform students that they are subject to a high-risk AI system. GDPR transparency duties already require the university to tell you about automated decision-making and its logic.

Can I get an explanation of how the AI contributed to a decision about me?
Yes — Article 86 gives you the right to obtain clear and meaningful explanations of the role of the AI system in the decision-making procedure and the main elements of the decision. The timing question (2 August 2026 vs 2 December 2027) is not yet settled, but you can invoke it now.

What if a human reviewed the AI's flag?
Then GDPR Article 22 (solely automated decisions) does not apply — but Article 86 can still apply if the system is high-risk. A rubber-stamp approval does not count as meaningful human involvement.

What is the difference between Article 85 and Article 86?
Article 85 is the right to lodge a complaint with a market surveillance authority (any person, any infringement). Article 86 is the right to an explanation from the deployer. Do not conflate them.

Can I read the university's Fundamental Rights Impact Assessment?
No — the FRIA is not published to students; its results go to the market surveillance authority. Ask whether one was done, and use GDPR access and transparency rights for the underlying data.

Where do I complain if the university ignores me?
First the Examination Board (watch the deadline). Then the national data protection authority (GDPR, Article 77) for the data side, and the market surveillance authority (AI Act, Article 85) for the AI Act side.

The bottom line

Three things to remember. First: emotion recognition in education is banned since 2 February 2025 — if exam software reads your face to infer stress or attention, that feature is illegal, not merely unfair. Second: proctoring and grading are high-risk — from 2 December 2027 your university must provide human oversight, keep the logs for at least six months, tell you the system is in use, and assess the fundamental-rights impact before first use. Third: you can demand an explanation — Article 86 gives you the right to clear and meaningful explanations of the AI's role in a decision about you, and GDPR Article 22 gives you the right not to be subject to a decision based solely on automated processing, with human intervention and the right to contest. The rights are new and untested in court for exams — but they are on the books, and the first step is always the same: ask for the explanation in writing, keep your evidence, and meet the appeal deadline.

Related guides

  • The EU AI Act Explained for Students: What Actually Applies to You in 2026
  • GDPR and ChatGPT: What Students Should Never Paste Into AI (and How to Opt Out in 2 Minutes)
  • Dutch Universities Ban AI Detectors: What It Means for You

Facts checked in November 2026 against Regulation (EU) 2024/1689, consolidated text 27 July 2026 (EUR-Lex, CELEX 02024R1689-20260727) — Arts. 3(39), 5, 6, 13, 14, 26, 27, 46, 49, 50, 77, 79, 85, 86, 87, 99, 111, 113; Annex III point 3; Recitals 44, 96, 171 — plus Regulation (EU) 2026/1744 (Digital Omnibus on AI), Commission guidelines on prohibited AI practices C(2025) 5052 (29 July 2025), Regulation (EU) 2016/679 (GDPR) Arts. 6(1)(e), 13(2)(f), 14(2)(g), 15(1)(h), 22, 35(3)(a), 77, CJEU SCHUFA C-634/21 (7 December 2023), and the Amsterdam District Court UvA proctoring case (12 June 2020).