glossaryPage.heroH1

Live proctoring has a human proctor from the vendor watching the student in real time on webcam. Record-and-review captures the session and flags events for human review by the institution or vendor after the exam. Automated proctoring uses ML models to flag events with no human review of unflagged sessions and only spot-check review of flagged sessions. Live proctoring is highest cost per exam and lowest false-flag rate; automated proctoring is lowest cost and highest false-flag rate. Most institutions use record-and-review as a middle ground.

Mixed evidence. Proctoring software does deter casual cheating (the presence of a webcam and lockdown browser changes student behaviour). It is less effective against determined cheating that uses a second device out of camera view, a coordinated stand-in test taker on a different computer, or contract cheating services that answer for the student. Research from the University of Colorado and NEPC finds that assessment redesign (open-book, oral defence, portfolio, in-person high-stakes tests) is often more effective than proctoring at ensuring assessment integrity. Many institutions now use proctoring for moderate-stakes remote exams and in-person testing for high-stakes summative exams.

Independent studies at multiple US universities have documented that ML-based face-detection and behaviour-flagging in some proctoring products has higher false-flag rates for students of colour, students with dark skin tones, students wearing religious head coverings, students with disabilities affecting eye movement or motor control, and students with unstable internet access. NEPC (National Education Policy Center) research and the EFF (Electronic Frontier Foundation) have both published critical reviews. The EU AI Act 2024 requires bias audits for high-risk education AI, which includes automated proctoring. Institutions should demand transparent bias reporting from vendors and provide alternatives (in-person testing, oral defence) for students who prefer or need them.

Under FERPA in the US, proctoring vendors that access education records fall under the school official exception, which requires direct institutional control and prohibits secondary use of the data. Under GDPR in the EU, the vendor is a data processor with a Data Processing Agreement required, and biometric data (face capture) is a special category requiring explicit consent and stronger safeguards. Some EU data protection authorities have issued warnings against specific proctoring products under GDPR. Institutions should publish clear notices to students on what is captured, how long it is retained, and how to request review or deletion.

OpenEduCat does not build its own proctoring product. The openeducat_exam and openeducat_lms modules integrate with external proctoring vendors (Respondus Monitor, ProctorU, Honorlock, Examity) via LTI and standard exam APIs for institutions that require remote proctoring. Because OpenEduCat is open source under LGPLv3 for the community edition, institutions retain full control of the exam data path and can choose the proctoring vendor whose privacy terms, bias reporting, and equity practice best match their policy.