The modern cheating flow for online exams no longer involves a second monitor — it involves a second operating system. A student runs the exam inside an Android emulator on a desktop, with an automation script answering questions and a wiki open in the host browser behind it; or takes the exam on a rooted phone where screen-sharing and answer-injection tools operate with system privileges the exam app cannot see. Browser-based proctoring is blind to all of this, because from inside the browser every environment looks like a laptop in a bedroom. Real exam security has to interrogate the environment itself: is this a real device, an unmodified operating system, a session without virtual machines and sandboxes around it? That interrogation — emulator detection, root and jailbreak detection, VM and sandbox detection — is the wall the student never sees and cannot argue with.

The four environments that invalidate an exam #

  • Android and iOS emulators: the "phone" is a desktop window with automation tooling one hotkey away.
  • Rooted Android devices: system-level privileges let answer overlays and remote control hide from the exam app.
  • Jailbroken iPhones: bypassed OS restrictions enable screen capture and background apps the exam forbids.
  • Virtual machines and desktop sandboxes: a controlled bubble where nothing proctored can be trusted.

Blocking, logging, alerting — in that order #

The response sequence matters as much as the detection. Blocking is the default: an emulator session simply cannot start the exam, which converts cheating attempts into study sessions. Logging catches the sophisticated cases — suspicious access attempts are recorded and alert the administrator automatically, so the academy sees the pattern (same student, three emulator attempts before the real device) that a single blocked session hides. Alerting closes the loop with humans: the administrator gets the signal with context, not a raw event, and can intervene before results are certified. This trio is what separates exam integrity from exam theater — a locked-down browser session with no environment checks produces confidence in the dashboard and none in the diploma.

Institutions evaluating this layer should ask vendors two questions. First: "Which environments do you refuse, and where is the refusal enforced?" — client-side checks that merely report home are advisory, not protective. Second: "What does the administrator actually receive?" — a list of blocked device models is noise; a per-student timeline of attempts with environment details is evidence. Randomised question banks, timed sessions, and auto-grading finish the picture, but they all stand on the same foundation: the exam happens on a real device, and everything else is secondary to that fact.