Lockdown Browser Alternatives for the AI Cheating Era

A lockdown browser is a restricted application that takes over the exam window: it blocks new tabs, disables copy-paste, suppresses notifications, and prevents navigation away from the assessment. Respondus LockDown Browser, Safe Exam Browser, and ExamSoft Examplify are dedicated lockdown browsers; Honorlock and Proctorio ship secure-browser modes that do the same thing inside the browser.

No. An invisible overlay like Cluely is a separate desktop application that marks itself invisible to screen-capture APIs (WDA_EXCLUDEFROMCAPTURE). It never touches the locked browser window, so no lockdown browser, by architecture, can see it. This is a layer boundary, not a detection gap.

No. Local LLMs run entirely in device memory with no outbound network traffic. A lockdown browser has no visibility into processes outside its own window, so it cannot detect, flag, or block local AI inference.

It solves it for exactly as long as the binary's name stays the same, which is hours, in practice. A denylist entry is a string match against a process name or bundle ID; renaming the executable, recompiling from source, or forking under a new name clears it instantly with zero code changes. The four invariants Aiseptor checks instead (capture-exclusion flags, renderer-child process trees, install-path fingerprints, and network-layer handshake fingerprints) are properties of how the technique works, not what the tool is called, so they don't reset on a rename.

No. Cluely is an OS-layer overlay that marks itself invisible to screen-capture APIs. Respondus operates at the browser layer and cannot see processes running outside its own window, and Respondus Monitor's recording is made through the same capture APIs the overlay excludes itself from. This is an architectural limit, not a detection gap: no browser-layer tool can see an OS-layer overlay.

No. LockDown Browser locks down the testing environment; it does not record you. Respondus Monitor is a separate application that adds webcam recording on top of it, and Respondus describes it that way itself: Monitor builds upon LockDown Browser. Monitor is automated rather than live. It records the session and surfaces flagged events for the instructor afterwards. Whether you are recorded therefore depends on whether your institution licensed Monitor and enabled it for that exam.

Only if the camera can see it. Respondus Monitor flags a device detected in a student's hand during an online exam, and says the system is trained to identify one even when only a small part is visible in the video. That is computer vision on the webcam feed. Respondus publishes no network, Bluetooth or proximity detection, so a phone resting outside the camera's frame is outside what the product claims to catch. This is the second-device gap network-layer enforcement is built for.

No. Respondus states that copying and pasting to or from an assessment is prevented, printing and screen capture are disabled, right-click menus, function keys, keyboard shortcuts and task switching are disabled, and the assessment runs full-screen and cannot be exited until submitted. It also blocks other applications including messaging, screen-sharing, virtual machines and remote desktops. The lockdown inside the exam window is genuinely strong; what it does not govern is a second device beside it, or a process that started before it did.

Both, with conditions. On ChromeOS it is a browser extension rather than the native application, and it supports exams delivered through Canvas, Blackboard Learn, D2L Brightspace, Schoology and Moodle. On iPad the instructor has to switch on an optional setting before a webcam exam will run, and Respondus notes that screen recording is not supported on iPad at all. Windows 10 and 11 and macOS 12 or later run the full application. Aiseptor runs on Windows and macOS without managed-device enrolment.

There isn't a single drop-in replacement: the honest framing is coverage, not substitution. Keep the lockdown browser (or move to Aiseptor Secure Browser, see above) for tab and window control, and add network-layer enforcement underneath it for the surface no browser-layer tool reaches. Institutions running both report the browser layer catching casual tab-switching while the network layer catches the overlay and local-LLM cases that would otherwise show a clean session log.