Catch cheating. Never punish the honest.
AI proctoring that flags genuine malpractice with surgical precision — without creating a surveillance nightmare for the 98% of candidates who are playing fair.
8 signals. Monitored simultaneously. Every second.
Our AI runs entirely on the candidate's device — no video streamed to the cloud — making it fast, private, and effective even on slow connections.
Multiple Face Detection
Flags when more than one person enters the camera frame. Distinguishes between a second person and a reflection. Timestamps every event.
Gaze & Eye Tracking
Detects when the candidate looks away from the screen repeatedly. Configurable threshold — e.g. flag after 3 off-screen looks of 3+ seconds.
Tab Switch & Focus Loss
Every window switch, minimise, or loss of browser focus is logged with a timestamp. Configurable auto-submit after N violations.
Audio Anomaly Detection
Detects voices other than the candidate speaking. Useful for catching live assistance. Loud ambient noise is distinguished from directed speech.
Mobile / Device Detection
Computer vision scans for smartphone shapes in the camera frame. Flags candidates holding or glancing at a second device.
Copy-Paste & Keyboard
All keyboard shortcuts blocked. Copy-paste, print-screen, developer console, and task-switch hotkeys are disabled at browser level.
Full-Screen Enforcement
Candidate must remain in full-screen mode. Exiting pauses the exam and logs a violation. Third violation can trigger auto-submit.
Absent Candidate Detection
Flags when no face is detected in the camera frame for more than a configurable threshold (default: 15 seconds).
Every flag. Full context. One click to review.
After the exam, you get a prioritised incident report sorted by severity. Click any flag to jump to the exact video timestamp. Review only what matters — not hours of footage.
- Flags sorted by severity — review high-risk candidates first
- Click to jump to exact video frame for any flag
- Export incident report as PDF for disciplinary records
- Mark flags as Genuine Violation or Dismissed
- Bulk-dismiss low-severity flags with one click
Privacy by design
- All AI processing runs on the candidate's device — no video is streamed or stored on ExamRankers servers
- Webcam footage (if recorded) is stored encrypted and accessible only to authorised institution admins
- Candidates are informed of monitoring before the exam starts — no hidden surveillance
- Video retained for 6 months then deleted; flagged events retained up to 3 years
- DPDPA 2023 compliant — full data processing agreement available
Fairness by default
- AI flags are evidence for human review — not automatic penalties
- Accommodation mode for candidates with disabilities — extended time, no gaze tracking
- Low-confidence flags are auto-marked for dismissal to reduce noise
- Discrimination impact testing run quarterly across gender, age, and skin-tone groups
- Candidates can submit a formal objection to any proctoring flag
Configurable sensitivity
- Set minimum duration before a gaze event is flagged (e.g. only flag if away for 5+ seconds)
- Configure max tab switches before auto-submit (1, 3, 5, or unlimited)
- Choose which signals to enable per exam — not all exams need full proctoring
- Set different thresholds for different exam types: mock tests vs. final exams
- Pilot mode — run AI proctoring silently to calibrate before enforcement
Beyond the basics
Live Video Proctoring
Pair AI proctoring with a live human invigilator who watches video feeds in real time. Available on Enterprise. Ideal for high-stakes board or professional exams.
Invigilator Dashboard
Supervisors see a live grid of all candidate feeds, AI flags, and connection status. Escalate or message a candidate directly from the dashboard.
Mobile App Proctoring
Candidates taking exams on the ExamRankers mobile app receive the same AI proctoring as desktop — face detection, audio analysis, and lock mode.
Geofencing & IP Lock
Restrict exam access to a physical location (GPS) or IP range. No VPN bypass — we detect and block common VPN providers automatically.
Identity Verification
Aadhaar-based e-KYC and photo match at login confirm candidate identity before the exam begins. Biometric terminal integration for CBT halls.
Risk Score per Candidate
Each candidate receives an AI-generated risk score (0–100) based on the number, severity, and pattern of flags. Sort candidates by risk for review.
AI Proctoring by plan
Basic security on all plans. Full AI proctoring from Professional.
| Feature | Starter | Professional | Enterprise |
|---|---|---|---|
| Browser lockdown | |||
| Tab-switch detection & logging | |||
| Copy-paste & shortcut prevention | |||
| Full-screen enforcement | |||
| AI face detection | |||
| Multiple face detection | |||
| Gaze / eye tracking | |||
| Audio anomaly detection | |||
| Mobile / device detection | |||
| Absent candidate detection | |||
| Proctoring incident reports | |||
| AI risk score per candidate | |||
| Configurable sensitivity thresholds | |||
| Accommodation / disability mode | |||
| Live video proctoring (human) | |||
| Live invigilator dashboard | |||
| Geofencing & IP lock | |||
| Aadhaar / biometric identity check | |||
| Video archive (6 months) | |||
| Bulk flag export for audit |
"We were sceptical about AI proctoring — worried about false positives upsetting genuine candidates. The pilot run on 500 students showed a 1.8% false positive rate, all in the 'Low' severity category. Not a single genuine complaint. We rolled it out to 50,000 the next month."
Run your next exam with complete confidence
Try AI proctoring free for 30 days. Enable it on any exam in under 60 seconds — no setup, no hardware.