Does TestGorilla Detect Cheating and AI? The 2026 Breakdown

TestGorilla detects cheating by watching browser-level behavior, not by seeing your screen or your other apps. It can take webcam snapshots about every 30 seconds, log full-screen exits, tab switches, and the mouse leaving the test window, record your IP address and location, block copy-paste, and block known AI agents. It groups those signals into behavior tiers for the employer, but it cannot see ChatGPT on a phone or in an app outside the browser.

Key Takeaways

• TestGorilla flags behavior, not intent. Its own position is that it does not label candidates as cheaters; it reports signals and lets the employer judge.
• The core signals are browser events. Full-screen exits, tab switches, mouse leaving the window, and opened developer tools are the most frequently logged actions.
• Webcam monitoring is periodic snapshots. Roughly one image every 30 seconds when enabled, with the candidate's permission, not continuous video.
• Prevention does much of the work. Copy-paste blocking, strict time limits, randomized questions from large pools, and question retirement make cheating slower and less useful.
• Behavior tiers summarize risk. Candidates land in No Issues, Some Issues, or Major Issues depending on how many signals fire and how severe they are.
• Anything outside the browser is invisible. Second devices, other apps, and second monitors are not observed, apart from what a snapshot happens to catch.

What Is TestGorilla and How Is It Different From Coding Platforms?

TestGorilla is a pre-employment testing platform that lets employers build assessments from a library of skills tests: cognitive ability, personality and culture fit, language, role-specific knowledge, software skills, and programming. It is used across many functions, not only engineering, which is the key difference from developer-focused tools like HackerRank or Coderbyte.

That breadth shapes its anti-cheating design. Most TestGorilla tests are short, timed, multiple-choice or short-answer modules, so the platform leans heavily on prevention (time pressure, copy-paste blocking, randomization) and on simple attention signals, rather than on code similarity analysis. If you are taking a TestGorilla coding test, expect the same proctoring as any other module, not a dedicated code plagiarism engine of the kind Codility or HackerRank advertise.

Which Anti-Cheating Measures Does TestGorilla Use?

TestGorilla's anti-cheating measures fall into three groups: monitoring signals that are logged with timestamps, prevention settings that make cheating harder, and identity and access controls. The table below lists them with what each one can and cannot tell an employer.

Employers choose which of these to enable. A candidate may see a test with every option on, or a lightweight cognitive test with only full-screen and tab tracking. The consent screen at the start tells you which monitoring is active.

How Do TestGorilla Webcam Snapshots Work?

Webcam monitoring on TestGorilla is snapshot-based. With the candidate's permission, the platform captures an image roughly every 30 seconds throughout the assessment. Snapshots serve two purposes: verifying that the invited candidate is the one taking the test, and giving the employer a visual record if other signals look suspicious.

A 30-second interval means a one-hour assessment produces roughly 120 images. That is enough to catch sustained patterns, such as a second person sitting beside the candidate or the candidate repeatedly looking down at something, but not enough to reconstruct every moment. A reviewer typically scrolls the snapshot strip only after another signal, such as a burst of full-screen exits, prompts a closer look.

Disabling the camera mid-test is a strong negative signal on its own. TestGorilla's behavior tiers treat a disabled webcam as one of the indicators that can push a candidate into the Major Issues tier. For more on how snapshot proctoring compares with continuous video across vendors, see our guide to webcam proctoring software.

Does TestGorilla Track Full-Screen Mode and Mouse Movement?

Yes. Full-screen tracking and mouse tracking are the two attention signals TestGorilla relies on most.

Full-screen mode tracking asks the candidate to take the test in full-screen and logs every exit. Pressing Escape, alt-tabbing to another app, or clicking a notification that pulls the window out of full-screen all register as exits. Because full-screen covers the whole display, an exit is a reasonable proxy for "the candidate looked at something else."

Mouse tracking checks whether the cursor stays inside the assessment window. A cursor that repeatedly leaves the test area suggests the candidate is interacting with something else on screen, such as a second window on a dual-monitor setup. This signal is coarse: it cannot see where the cursor went, only that it left.

Tab switching and developer tools detection round out this layer. Opening dev tools is notable because it is a classic route to inspecting page data, so TestGorilla logs it even though most candidates who do it are simply curious or clicked the wrong shortcut.

Can TestGorilla Detect ChatGPT and AI Tools?

TestGorilla cannot see ChatGPT directly. Like every browser-based assessment, it has no access to the list of applications on your computer and cannot look at your phone. What it does instead is make AI use slower, riskier, and more visible:

1. Copy-paste disabling removes the fastest workflow: copy the question, paste it into a chatbot, paste the answer back.
2. Full-screen and tab tracking log the moment you leave the test to consult an AI tool on the same machine.
3. Mouse tracking catches the cursor drifting to a second window.
4. Webcam snapshots may capture you looking down at a phone or away at another screen.
5. Time limits leave little room to type a question into a chatbot, wait for an answer, and evaluate it.
6. AI agent blocking, part of what TestGorilla calls its Trust Layer, aims to stop known AI agents and automation tools from accessing the assessment environment at all.

The Trust Layer is TestGorilla's umbrella term for its integrity safeguards, including identity verification up front, agent blocking, and turning patterns such as repeated tab changes and full-screen exits into structured insights for reviewers. TestGorilla also positions it as keeping completion rates above 90 percent, which signals the trade-off the company is making: enough friction to deter cheating, not so much that honest candidates abandon the test.

What none of these layers can do is detect an answer that was read from a second device and typed in by hand, or knowledge that the candidate looked up before the test. That is the structural limit of browser-level proctoring, and it is why TestGorilla's own messaging emphasizes prevention and human review over automated verdicts.

What Are TestGorilla Behavior Tiers?

Behavior tiers are TestGorilla's summary layer. Instead of handing employers a raw log of every tab switch, TestGorilla groups each completed assessment into one of three risk levels based on the depth, intensity, and variety of flagged behavior.

The tier is not a cheating verdict. TestGorilla explicitly says it does not label candidates as cheaters because proving intent is impossible, and it does not automatically rule out candidates based on suspicious activity. The tier is a prompt for human judgment.

What Are the Limits of TestGorilla's Cheating Detection?

TestGorilla's detection has clear limits, and understanding them protects honest candidates as much as anyone else.

What it structurally cannot see

• Other applications. It cannot list processes or read windows outside the browser.
• Second monitors. It sees the cursor leave the window, not what is on the other screen.
• Second devices. A phone or tablet is invisible unless a snapshot happens to catch you looking at it.
• Audio. Snapshot proctoring does not analyze what is said in the room.
• Prior knowledge. If questions have leaked and been memorized, nothing in the session reveals that, which is why randomization and retirement exist.

Where false positives come from

Behavior signals are noisy. Common innocent causes include:

1. Operating system notifications or update prompts that break full-screen mode
2. Dual-monitor setups where the cursor drifts across the edge of the display
3. Laptop webcams that drop out due to driver or power-saving issues
4. Accidentally pressing a keyboard shortcut that opens developer tools
5. VPNs, corporate networks, or mobile hotspots that report a surprising location

If any of these happen, tell the recruiter promptly and specifically. A short note such as "my camera disconnected around minute 20 and I restarted it" carries real weight when a reviewer later sees the gap in the snapshot strip. Our guide on how to pass a proctored test has a full pre-test checklist.

How Does TestGorilla Compare With Other Assessment Platforms?

TestGorilla sits at the general-purpose end of the market. Coding-first platforms put more effort into analyzing the code itself, while TestGorilla puts more effort into controlling the testing environment across many test types.

If your hiring process moves from TestGorilla into coding-specific rounds, the detection model changes. Read does Codility detect AI and can interviewers see your screen during a coding interview to understand what comes next.

How Should You Prepare for a TestGorilla Assessment?

Treat a TestGorilla test as a short, monitored exam where clean behavior matters as much as the score.

• Close every app that sends notifications, and silence your phone.

• Use one monitor if you can, or keep your cursor firmly on the test display.

• Check your webcam and browser permissions before starting, using the same browser you will test in.

• Stay in full-screen for the entire test and avoid keyboard shortcuts you do not normally use.

• Read each question once, carefully. Time limits are tight by design.

• Report technical interruptions in writing right after the test.

• If the test includes a coding module, write your answer directly in the provided field. With copy-paste disabled, drafting elsewhere only wastes time.

• Do not take the same assessment twice from the same network under a different email. IP logging is designed to catch exactly that.

It also helps to remember what a reviewer actually sees. They do not watch you live. They open a report with your score, your behavior tier, a timestamped list of flagged events, and, if enabled, a strip of webcam snapshots. A clean report with a solid score rarely gets a second look. A strong score paired with a Major Issues tier gets scrutinized, and that scrutiny often decides whether you move forward.

The skills test is usually a filter rather than the decision. Candidates who pass are typically invited to live interviews, and that is where most of the real evaluation happens.