How Respondus and Lockdown Browser Pair With AI Detection in Canvas

**Respondus LockDown Browser and AI detection solve opposite halves of the same problem: LockDown Browser *prevents* AI access during a Canvas exam by sealing the test environment, while AI detection *examines* what was submitted afterward — and Respondus Monitor sits between them, watching the room by webcam.** Searches for “respondus lockdown browser ai detection” usually come from students (and sometimes instructors) who’ve blurred the layers into one all-seeing system. It isn’t one system. It’s a stack, each layer with a specific job and specific blind spots, and both sides of the exam benefit from knowing which layer does what.

Key takeaways

• LockDown Browser is prevention: it seals the browser during the exam and analyzes nothing you write.
• Respondus Monitor is observation: webcam recording with automated flags for review-worthy moments, scored for instructor attention.
• Text-level AI detection (typically Turnitin in Canvas) is post-hoc analysis of the submitted answers themselves.
• The stack’s blind spots are physical — second devices, off-camera helpers — which is why the layers exist together.
• For honest students, the layers are mostly protective: a clean proctoring record is powerful evidence against a false text flag.

Layer one: what LockDown Browser actually does

When an instructor enables LockDown Browser for a Canvas quiz, students must take that quiz in Respondus’s custom browser rather than Chrome or Safari. Inside it, the machine becomes single-purpose: no new tabs or windows, no switching to other applications, copy-paste disabled, printing and screen capture blocked, right-click menus gone, and the exam locked to full screen until submission.

Notice what’s absent from that list: any analysis of your answers. LockDown Browser never reads your prose, computes no AI probability, and wouldn’t care if you typed a ChatGPT-style essay from memory. Its entire theory is environmental — if the machine can’t reach ChatGPT, the paste can’t happen on that machine. The qualifier matters, and Respondus knows it, which is why there’s a layer two.

Layer two: Respondus Monitor’s AI flags

Monitor is the webcam add-on that runs on top of LockDown Browser for remote exams. It records the student throughout the session and runs automated analysis on the video — this is the only place “AI” appears on the Respondus side, and it’s computer vision, not text analysis. The system flags events like the student leaving the frame, an additional face entering it, the face being obscured or persistently turned away, and interruptions in the recording, then aggregates them into a review priority that tells the instructor which sessions deserve human eyes first.

Two properties keep this layer honest. Flags are review prompts, not verdicts — Respondus’s own materials frame them as starting points, and a flagged session usually contains something as mundane as a roommate walking past. And the deterrence is most of the product: a student deciding whether to prop a phone against the monitor behaves differently when a camera is rolling and glance-patterns are analyzed.

Layer three: AI detection on the submitted text

The last layer runs after the exam ends and has nothing to do with Respondus at all. Essay-style responses in Canvas can flow through Turnitin, where AI writing detection scores eligible submissions — we’ve detailed how a Turnitin AI flag surfaces in Canvas SpeedGrader from the instructor’s seat. This layer doesn’t care how sealed the exam environment was; it reads the words themselves. A student who transcribed answers from a phone past every proctoring control still hands this layer exactly the evidence it analyzes.

Timed exam prose deserves a caveat here, and it’s an important one: students writing under pressure produce flatter, more formulaic sentences than they would at home, and statistical detectors read flatness as machine-likeness. We’ve written about AI detectors and timed essay exams precisely because the false-positive risk runs *higher* in this genre. If you want intuition for what “flat prose” means statistically, our free AI detector shows the sentence-level heatmap — exam answers written in a hurry light up more than people expect.

This is also where the stack turns protective for honest students. A text flag alone is ambiguous. A text flag on an exam taken in LockDown Browser, with a clean Monitor recording showing forty minutes of a student visibly composing at a keyboard? The proctoring layers just became the accused student’s best exculpatory evidence. The layers check each other in both directions.

Where Respondus LockDown Browser AI detection still fails

The documented blind spots are physical-world problems: a phone below camera line, notes taped behind the laptop, a helper outside the frame, an entire second computer. Monitor’s glance-detection catches some of this; determined circumvention still happens. That residual gap explains two things about how institutions behave — why they keep the text-analysis layer even for proctored exams, and why integrity offices build cases from converging evidence (environment, video, text, voice-match with prior work) rather than any single flag.

For students, the strategy that dominates every alternative is the boring one: treat proctored exams as the closed-book events they are, set up your room honestly, and write in your own voice with course-specific details — the voice-match with your other work is itself evidence in your favor. Where AI use is actually permitted — take-home essays, drafting, revision — the rules and tools are different, and that’s the territory our humanizer serves (see pricing); the blog maps which situations are which. Knowing what each layer sees isn’t a cheating manual. It’s the difference between fearing an all-seeing system that doesn’t exist and respecting the specific, fallible, mutually reinforcing one that does.

Frequently asked questions

Does Respondus LockDown Browser detect AI like ChatGPT?

Not in the text-analysis sense — LockDown Browser is prevention, not detection. It locks the exam environment so ChatGPT can’t be opened on that computer: no new tabs, no other applications, copy-paste and printing disabled. It doesn’t analyze your answers for AI-generated writing at all. Text-level AI detection, if your institution uses it, happens separately — typically through Turnitin scanning essay responses after submission. The two address the same worry from opposite ends.

What does Respondus Monitor’s AI actually flag?

Monitor’s automated review works on the webcam video, not your prose. It flags proctoring events — the student leaving the frame, a second person appearing, the face not matching or being obscured, extended looking away from the screen — and rolls them into a review priority so instructors know which recordings deserve human attention. Respondus positions these as flags for review rather than accusations, and an instructor is supposed to watch the footage before concluding anything.

Can I still get flagged for AI if the exam used LockDown Browser?

Yes, in two ways. If you found a path to AI text anyway — a phone, a second computer — the pasted or transcribed answer can still be caught by Turnitin’s AI scan or by the mismatch with your known writing voice; LockDown Browser’s presence doesn’t immunize the submission. And false flags remain possible too: a student who writes formal, evenly paced prose under exam pressure can trip a text detector even in a fully proctored session, which is exactly when proctoring records ironically become their best defense.

What can the LockDown Browser and Monitor combination still not see?

Its documented blind spots are physical: a phone propped below camera view, notes on the wall behind the laptop, a second person outside the frame feeding answers, or a whole second computer. Monitor’s webcam flags catch some of this — repeated off-screen glances are a classic flag — but determined circumvention happens, which is why integrity offices treat proctoring as deterrence plus evidence rather than a guarantee, and why the text of the answers still gets human and algorithmic scrutiny afterward.

As a student, how should I prepare for a proctored Canvas exam with AI checks?

Treat it as closed-book unless the instructor says otherwise, and set up honestly: clear desk, working webcam, stable connection, no second devices in reach. If tech problems hit mid-exam, email the instructor immediately with timestamps — documented glitches read very differently from unexplained gaps. And don’t over-sanitize your writing out of detector fear; answer in your natural voice with course-specific details. Your in-class voice showing up in the exam is precisely what makes your work believable.