Partly, and TestGorilla is unusually honest about the "partly." It takes webcam snapshots roughly every 30 seconds, watches for full-screen exits, tab switches, developer tools, pasting and IP mismatches, then sorts everything into a three-tier behavior rating instead of labeling anyone a cheater. It does not install anything on your machine, does not lock down your operating system, and does not scan your running processes. In TestGorilla's own words, the system "doesn't and cannot tell you who cheated; it can point to behaviours which you can contextualise."
Last checked: September 17, 2026. Every proctoring claim below links to TestGorilla's own documentation. We build InterviewMan, an AI interview assistant, so read the practical advice with that in mind.
Key takeaways:
- TestGorilla's anti-cheating measures include webcam snapshots about every 30 seconds, full-screen exit detection, tab switching, developer tools logging, IP and approximate location, and paste tracking.
- Results are grouped into three behavior tiers rather than a single cheat verdict, because TestGorilla treats cheating as "a spectrum" and wants to avoid punishing a single stray click.
- The company publicly states that "no assessment has ever been 100% cheat-proof" and that "None of this is magic. None of it is a guarantee."
- Everything TestGorilla observes is observable from inside a web page. There is no lockdown browser, no installed agent and no documented process scanning.
- The snapshot camera is the sharpest tool in the set. A candidate reading answers off a second screen produces a very legible head-turn pattern every 30 seconds.
- TestGorilla's stated strategy is shifting away from detection and toward assessment design, which is genuinely a better answer than most vendors give.
What TestGorilla is and who uses it
TestGorilla is a general skills-assessment platform, not a coding-only one. Employers assemble a battery of short tests: cognitive ability, personality, language, software-specific skills, role-specific knowledge, plus optional coding tests and one-way video questions. Small and mid-sized companies use it heavily for non-engineering roles, which is why so many candidates meet TestGorilla for a customer support, marketing, finance or operations job rather than a software one.
That shapes the detection model. A one-hour multiple-choice battery does not need a lockdown browser to be useful. It needs enough signal to tell a reviewer when to look twice. TestGorilla built exactly that.
If your assessment includes a coding test, the report has its own structure, documented in interpreting a coding test report.
Ready to ace your next interview?
InterviewMan gives you real-time AI answers during live interviews — undetectable on Zoom, Meet, and Teams.
Try InterviewMan FreeWhat TestGorilla monitors
Webcam snapshots, roughly every 30 seconds
TestGorilla captures still images of the candidate at a regular interval, documented at about every 30 seconds, and attaches them to the report for identity verification. In TestGorilla's own overview of anti-cheating tools, the feature is listed as "Regular webcam snapshots of candidates (with candidates' permission)."
This is the only measure that can be switched off, and it is also the most informative one for a reviewer. Snapshots are not video, so a single glance away is invisible. A repeated head turn in the same direction across a dozen snapshots is not.
Full-screen mode
The system monitors whether you stay in the full-screen test window. Exiting full-screen is recorded as a distinct event with a count.
Tab switching
Leaving the assessment tab is logged. Unlike HackerEarth, TestGorilla's documented behavior is to record the event and feed it into the behavior tier rather than to log you out at a fixed threshold.
Developer tools
Opening browser developer tools during an assessment is logged. This one is worth spelling out because candidates sometimes open the console out of curiosity or to fix a rendering problem, and it lands in the report as a flag.
IP address and approximate location
TestGorilla records the IP address and approximate location, and uses repeat IPs to identify multiple attempts from the same source, which can indicate test sharing or one person sitting several assessments.
Paste tracking and copy-paste controls
Pasting into answer fields is tracked, and copy-paste can be disabled entirely on many test types.
Question randomization and server-side timers
TestGorilla draws from randomized question banks with exposure limits, and uses server-side timers that "keep running even when a window closes," so closing the tab to buy thinking time does not work. It also surfaces an honesty agreement at the point of submission.
Camera disabled or turned off
If the camera is switched off mid-assessment after being enabled, that state change is itself recorded.
What TestGorilla cannot see
Every signal in the list above is something a web page can observe about itself: its own visibility state, its own full-screen status, its own paste events, the browser's developer tools hooks, the connecting IP address and the camera stream you granted it.
What is not in TestGorilla's published documentation, anywhere: a lockdown browser, an installed proctoring agent, virtual machine detection, running-process enumeration, or extension enumeration on your operating system. TestGorilla is a web application, and its detection surface is the web application's own window.
The consequences follow directly:
- A native desktop application that is excluded from screen-capture APIs and never injects into the assessment page is not something the page can query. It is outside the boundary.
- A second device, a phone or tablet, is outside the boundary by definition. Nothing on the test machine can see it.
- A second monitor is not detected as hardware.
And now the part that undoes half of that: the webcam. TestGorilla's snapshots exist precisely because a reviewer cannot see your desk, but they can see your face. Reading from a second screen or a phone produces a consistent gaze direction, and consistent gaze direction across twenty stills is the single most common reason a human reviewer opens a second look. Software invisibility does not make your eyes point at the monitor.
TestGorilla is also explicit about its own limits, which we should credit. Its article Why Cheat-Proof Hiring Assessments Simply Don't Exist argues that detection creates friction but cannot prevent determined cheating, and the companion piece says the quiet part out loud: the anti-cheating suite "doesn't and cannot tell you who cheated."
How TestGorilla scores or flags suspicious behavior
TestGorilla replaced a raw list of flags with what it calls behavior tiers. Individual signals are grouped into three levels: No Issues, Some Issues, and Major Issues.
Ready to ace your next interview?
InterviewMan gives you real-time AI answers during live interviews — undetectable on Zoom, Meet, and Teams.
Try InterviewMan FreeThe design intent, in TestGorilla's own framing, is to contextualize a candidate's actions instead of reading each flag in isolation, recognizing that "cheating is a spectrum" and distinguishing "the depth, intensity, frequency, and variety of concerning or non-concerning behaviors." The system deliberately allows for occasional non-concerning behavior, "since a single data point alone may not reliably indicate an intent to mislead."
In practice that means:
- One full-screen exit while a notification popped up: usually stays in the lowest tier.
- Several tab switches plus a paste event plus a gap in webcam snapshots: escalates.
- Multiple categories firing repeatedly across the assessment: Major Issues, and a reviewer will look.
This is a materially fairer system than a binary cheat flag, and it is also why gaming a single signal does not help much. The tier reacts to variety, not just to count.
What recruiters see in the report
The report shows your scores per test, the honesty agreement acknowledgement, the behavior tier, the individual anti-cheating signals with counts, the webcam snapshot gallery, and the IP and approximate location. Coding tests add code output, test-case results and, in TestGorilla's coding report, an analysis of the submitted solution.
Recruiters typically scan the tier first. A "No Issues" label means they read the scores and move on. "Major Issues" means they open the snapshots. That review order is the same across the industry, and we describe it in more detail in how companies detect AI cheating in interviews.
One nuance specific to TestGorilla: because the platform is often used for non-technical roles, the reviewer opening your report may be a hiring manager rather than an engineer. They tend to trust the tier label more, not less.
Using an AI interview assistant on TestGorilla: what works, what is risky
What is technically true. InterviewMan is a native desktop app rather than a browser extension. Its overlay is excluded from screen capture and recording APIs, it is hidden from the dock, the taskbar and the app switcher, and it never injects anything into the assessment page. The full documented list is on our undetectability page. Against a purely browser-scoped detection surface like TestGorilla's, that architecture is doing real work: there is no tab to switch, no extension to enumerate, no paste event to log.
What is risky here specifically. The webcam. Snapshots every 30 seconds, running for the whole assessment, produce a gaze record. If your eyes travel to a phone beside your keyboard or a second monitor to your left, the pattern shows. This is the single biggest reason we tell people that on TestGorilla the placement of your help matters more than its software stealth. An on-screen overlay positioned near where you would naturally look beats a second device sitting off to one side.
What breaks immediately. Chrome extensions and browser overlays. They render inside the tab TestGorilla is watching, and the developer-tools and paste logging sit right next to them.
What we do not claim. InterviewMan does not hide network traffic, does not type or submit answers for you, does not execute code, and does not integrate with proctoring platform APIs. We report zero confirmed detections across 570,000+ users, which is a track record and not a technical guarantee. It is also no help at all against a snapshot of you looking sideways.
The policy question, honestly. TestGorilla surfaces an honesty agreement at submission. Clicking it and then using undisclosed assistance is a breach of the thing you just agreed to, and the employer may treat a Major Issues tier as sufficient grounds to end the process. TestGorilla also cannot prove it, which is the uncomfortable middle ground most of these decisions sit in.
Our full case on both sides is in Should you use AI in a job interview in 2026?, and the state of company policies is in which companies allow or ban AI in interviews.
Where an assistant is unambiguously fine. Preparation. A TestGorilla battery usually mixes cognitive tests, situational judgement and role-specific knowledge. That is all rehearsable.
How to prepare for a TestGorilla assessment honestly
- Find out which tests are in your battery. Recruiters will usually tell you if you ask. Cognitive ability, situational judgement and language tests are practiceable in a way that surprises people.
- Practice under time pressure. Most TestGorilla tests are 10 to 20 minutes each. Timed practice matters more than depth here.
- Set up your desk before you start. One screen, phone away, notifications off. This removes both the temptation and the gaze pattern.
- Do a full dry run. Camera on, full-screen, headphones. Our interview prep hub has a checklist.
- Rehearse the video questions out loud. Many TestGorilla batteries include one-way video answers. AI mock interview practice is the closest analogue, and the STAR method guide covers the structure reviewers expect.
- If there is a coding test, treat it as a real assessment. See our 10-step coding interview plan and technical interview preparation guide.
How TestGorilla compares to CodeSignal, HackerRank, CoderPad and Codility
| Platform | Lockdown app | Webcam | Screen capture | Tab switching | Dev tools logged | Scoring model | What usually flags you |
|---|---|---|---|---|---|---|---|
| TestGorilla | No | Snapshots about every 30 seconds | No full-display recording | Logged into behavior tier | Yes | Three behavior tiers | Repeated gaze direction, multiple signal types together |
| CodeSignal | No | Yes, plus ID verification | Screen and webcam recording in AI Proctoring | Logged | Not published | Suspicion Score | Edit history, paste events |
| HackerRank | No | Optional | Browser-level proctoring | Logged with timestamps | Yes | Flag list | Tab exits, clipboard, extension checks |
| CoderPad | No | Only via the video call | Interviewer watches the pad live | Not the main signal | Not published | Interviewer judgement plus playback | Paste events, typing playback |
| Codility | No | Optional | Browser events | Tracked | Not published | Similarity plus event log | Tab switching, similarity |
| HackerEarth | Yes, Smart Browser (optional) | Frequent snapshots | Records the test window | Auto-logout at threshold | Blocked | Violation counts | Threshold breaches, plagiarism |
| Vervoe | No | No webcam monitoring | Random screenshots | Not published | Not published | Plagiarism and geolocation flags | Text similarity, location mismatch |
TestGorilla is at the lighter end of this list on software control and at the honest end on how it talks about detection. Those two facts are related.
Frequently asked questions
Does TestGorilla use webcam monitoring?
Yes, it captures webcam snapshots at a regular interval, documented at roughly every 30 seconds, and attaches them to the report so reviewers can check identity and movement. It is the one anti-cheating measure that can be turned off, and it requires your permission.
Does TestGorilla detect ChatGPT?
Not by identifying the tool itself. It detects the behaviors that using it tends to produce: tab switches, full-screen exits, paste events into answer fields and gaps or gaze patterns in the webcam snapshots. TestGorilla is explicit that its measures point at behaviors, not at proof of cheating.
Can TestGorilla see my whole screen or other apps?
No. TestGorilla's documented measures are all browser-scoped: full-screen state, tab visibility, developer tools, paste events, IP address and camera snapshots. There is no lockdown browser, no installed agent and no published process scanning.
What are TestGorilla behavior tiers?
They are a three-level rating of your assessment behavior: No Issues, Some Issues, and Major Issues. Instead of showing raw flags in isolation, TestGorilla groups them by depth, intensity, frequency and variety, so an isolated event does not carry the same weight as several different signals repeating.
Will one tab switch fail me?
Almost certainly not. TestGorilla designed behavior tiers specifically to allow occasional non-concerning behavior, on the stated basis that a single data point may not reliably indicate intent to mislead. Repeated switches combined with other signals are what escalate the tier.
Does TestGorilla block copy and paste?
Paste events are tracked, and copy-paste can be disabled on many test types depending on the employer's configuration. Even when pasting is possible, the event is recorded and feeds into the behavior tier.
Can TestGorilla prove I cheated?
By its own account, no. TestGorilla states that its anti-cheating suite cannot tell you who cheated and can only point to behaviors that a human then has to interpret, and that no assessment has ever been fully cheat-proof.
Do browser extensions get caught on TestGorilla?
An extension that renders on the assessment page is inside the window being watched, and it sits next to the developer-tools and paste logging. It is the highest-risk category of tool on any browser-based platform, TestGorilla included.
See also
Ready to Ace Your Next Interview?
Join 570,000+ professionals using InterviewMan to get real-time AI assistance during their interviews.
