EU AI Act

AI flags. Humans decide. Candidates know.

The EU AI Act asks for human oversight, transparency and no emotion recognition in education and the workplace. Here is how QProctor is built around those ideas, and what stays with your institution.

For privacy, this animation was created with AI and shows no real people. Try the system yourself →

The regulation

What the AI Act means for exam monitoring

Regulation (EU) 2024/1689, the AI Act, sets rules by risk level. Three parts matter most for proctoring.

Article 5(1)(f)

No emotion recognition

AI systems that infer the emotions of people in education institutions and the workplace are prohibited, except for medical or safety reasons.

Article 14

Human oversight

High-risk AI systems must be designed so that people can oversee them effectively while they are in use.

Article 26

Duties of the deployer

The organisation using a high-risk system must use it as instructed, assign human oversight and inform people when the system is used in decisions about them.

High-risk status Depending on how it is used, AI that monitors candidates during tests in education can fall into the Act’s high-risk category. Whether and how the Act applies to your exams is a question for your legal team.

Our approach

How QProctor is designed

No automated decisions

AI flags moments and shows the evidence. It never fails, bans or grades a candidate. A person at your institution reviews and decides.

No emotion recognition

Gaze analysis looks at where the candidate is looking. Voice analysis listens for a second voice in the room. Neither tries to read feelings, and nothing predicts behaviour.

Informed candidates

Before the exam, during activation, candidates read the privacy notice, give explicit consent, test their device and can take a practice exam.

Visible, not hidden

When the AI proctor has a concern, the candidate sees an on-screen proctoring notice. Monitoring is never invisible to the candidate.

Explainable evidence

Every flag comes with the frame from that moment and the recording. The risk score (0–100) is built from documented event types and weights.

Informational score

The integrity score sent to your LMS sits next to the grade as information. It does not change the grade.

Human oversight in practice

Tools that keep people in charge

  • Sensitivity levels (Strict, Balanced, Tolerant) chosen by your institution
  • Live proctors can message, warn, lock the screen or end the exam; each action is their decision
  • Smart monitoring brings flagged candidates to the proctor’s attention instead of deciding on its own
  • Post-exam review mode: reviewers go through the flagged moments and decide
  • Time extensions and extra attempts for candidates who had technical problems
  • Trusted timestamps from TÜBİTAK on the evidence
Who does what

Your institution and QProctor

TopicYour institutionQProctor
Decisions about candidatesReviews evidence and decidesFlags and shows evidence; makes no decision
Informing candidatesDecides what candidates are told and whyShows notices and collects consent on your behalf during activation
Level of monitoringChooses which checks are on and the sensitivityOffers each check as a separate option
Human oversightAssigns proctors and reviewersProvides live and review tools
DataChooses region and retention periodProcesses data only on your instructions
Your assessmentCarries out its own legal and impact assessmentAnswers questions and provides product documentation

Not legal advice This page explains how QProctor works. It is not legal advice and does not replace your own assessment under the AI Act or the GDPR.

FAQ

AI Act questions

Does QProctor use emotion recognition?

No. Our analyses cover liveness, gaze direction, objects, face match, person count, ID reading, a second voice and lighting. None of them infer emotions.

Can the AI fail a candidate on its own?

No. AI flags moments and shows the evidence. Every decision about a candidate is made by a person at your institution.

Does the risk score change the grade in our LMS?

No. The integrity score is written next to the grade as information for the instructor. The grade stays as it is.

How are candidates told about AI monitoring?

Before the exam, during activation, candidates read the privacy notice and give explicit consent. During the exam, AI proctor warnings appear on their screen.

Can we reduce how much is monitored?

Yes. Each check is a separate option, and you choose the sensitivity level. Turn on only what the exam needs.

Run an exam where people make the call

14-day free trial with 20 candidates. No subscription.

Start free