Blog/Can Online Assessments Detect Cheating? What They Track

Can Online Assessments Detect Cheating? What They Track

Tab switches, copy-paste, webcam AI flags, code plagiarism and AI plugins: what online assessment platforms actually track, and how flags are reviewed.

Last updated: 23 October 2026 · By the Asuraa Team

Quick answer: Yes, online assessments can detect many forms of cheating by combining signals rather than one detector. As of October 2026, Mettl's help centre lists webcam AI flags for a mobile phone, another person and looking away, with screen and camera snapshots every 60 seconds. HackerEarth can block copy-paste and scores tab switches and plagiarism in an integrity index. Testlify detects browser AI plugins. Most flags go to a human reviewer.

Key takeaways

  • Online assessment platforms detect cheating by combining signals such as tab switches, copy-paste, webcam analysis, code similarity and identity checks.
  • Mercer Mettl captures screen and webcam snapshots every 60 seconds and lists seven AI flags, including mobile phone detected and another person.
  • HackerEarth's integrity index lowers a candidate's score for tab switches, webcam snapshot violations and plagiarism.
  • Testlify detects browser AI plugins, warns on the first detection and removes the candidate after the employer's limit, which it suggests setting at 2 to 3.
  • Most webcam and copy-paste flags are reviewed by a person, while exceeding a tab-switch limit can log out or terminate the test immediately.

Yes, online assessments can detect many forms of cheating, though not through one magic detector. Platforms combine several signals: tab and window switches, copy-paste activity, webcam snapshots analysed by AI, screen captures, audio, code similarity between candidates, browser AI plugins and identity checks. Each signal is weak on its own; together they give the employer a report that makes most cheating visible to a human reviewer.

This guide lists what HackerEarth, Mercer Mettl and Testlify say they track, based on their help centres as of October 2026. For a platform-by-platform overview of hiring tests in India, see our online assessments guide for India.

What signals do online assessment platforms track?

They track behaviour on your computer, what the webcam and microphone capture, and what your answers look like compared with others. The table maps each signal to the platform documentation that describes it.

SignalWhat is trackedPlatforms documenting it
Tab or window switchEvery time the test window loses focus, including app pop-upsHackerEarth (proctoring settings), Mettl (anti-cheating measures), Testlify
Copy-pasteBlocked in the code editor, or logged with timestamp and questionHackerEarth (proctoring settings), Mettl blocks copy-paste and print-screen (anti-cheating measures), Testlify logs it (copy-paste tracking)
Webcam snapshotsPhotos at random or fixed intervals; Mettl captures screen and camera every 60 secondsHackerEarth, Mettl
AI face analysisFace not visible, another person, looking away, mobile phone detectedMettl (anti-cheating measures), HackerEarth looking-away category (release notes), Testlify face counts (proctoring flags)
AudioHuman speech and number of speakersMettl, off by default (anti-cheating measures)
Code similarityPlagiarism feeds a single integrity scoreHackerEarth integrity index (release notes)
Browser AI plugins and assistant pop-upsAI extensions detected and logged; AI tool pop-ups flaggedTestlify (AI plugin restriction, Focus Guard)
Second screen, VM, virtual cameraLockdown browser refuses to run or ends the testHackerEarth Smart Browser (candidate guide), Mettl Secure Browser (MSB)
IdentityID photo, face capture, Aadhaar via DigiLockerMettl, HackerEarth (ID verification)
NetworkOnly approved IP addresses can open the testHackerEarth, Mettl

Not every test uses every signal. The employer chooses, so a campus aptitude round may log only tab switches while a remote SDE test uses webcam, Smart Browser and plagiarism checks together.

Can coding tests detect copied code or ChatGPT answers?

They can detect the behaviours around it, and some detect the copied code itself. Copy-paste into the editor can be disabled outright on HackerEarth or logged with timestamps on Testlify, so pasting a full solution from another window is either impossible or recorded. HackerEarth's integrity index includes plagiarism among the violations that lower a candidate's score.

For AI tools specifically, detection works through the setup rather than the text. Testlify can detect browser-based AI plugins: per its AI plugin article, the first detection shows a warning, each detection is logged, and you are removed once the employer's limit is crossed (Testlify suggests a limit of 2 to 3). Testlify's Focus Guard flags pop-ups from AI tools and suspicious keyboard shortcuts. HackerEarth's March 2024 release notes say its "looking away" category targets candidates using a phone below the desk or a second laptop to reach Google or ChatGPT.

Code that is correct is still not automatically trusted. Codility's candidate help page explains that solutions are run against many test cases for correctness and scalability and the results go to the employer, who decides the next step. Many employers follow up with a live technical interview, where you may be asked to explain your solution.

How does webcam proctoring decide something is suspicious?

AI marks specific events in the snapshots or video, and a human reviews the marked moments. Mettl's anti-cheating measures list seven AI flags: face not visible, another person, away from window, looking away, face partially visible, participant not present, and mobile phone detected. Mettl can also record the full session and run a 360-degree room check before the test.

Testlify's proctoring flags article gives an example of thresholds: no face, multiple faces or another person's face more than 3 times, any AI plugin use, or any external copy-paste produce a red "violation" flag, while minor issues such as background noise produce a yellow flag that calls for manual review of the recording.

Innocent things can trigger flags: poor lighting, a family member walking past, looking down at rough paper. That is why platforms route flags to human review rather than automatic rejection in most setups.

What happens if the platform flags you?

Usually the employer reviews the evidence and decides; termination is immediate only for hard limits like tab-switch counts. Outcomes fall into three groups:

  1. Immediate. Exceeding a tab-switch limit logs you out (HackerEarth) or terminates the test (Mettl, Testlify with Terminate on). A Mettl Secure Browser test ends if you connect a cable to share your screen.
  2. Reviewed. AI webcam flags, copy-paste logs and plagiarism appear in a report. The employer may accept the result, discard it or call you to explain.
  3. Later rounds. Many employers re-test the same skills in a live technical interview. A candidate who scored full marks but cannot explain the solution is easy to spot. Employers are also being advised on this; see our employer guide on stopping candidates from using ChatGPT in tests.

Being caught can also mean being barred from that employer's future drives, and campus placement cells may have their own penalties. Ask your placement cell if you are unsure.

What do most guides on detecting cheating in online tests get wrong?

Most guides fall into one of two extremes: "they cannot detect anything if you use a second device" or "AI will instantly catch and disqualify you". Both are wrong.

A second device avoids tab detection but not the webcam. Mettl's mobile-phone and looking-away flags and HackerEarth's looking-away category were built for exactly that case. At the other extreme, platforms describe flags as evidence for a reviewer, not automatic verdicts, and Testlify explicitly recommends manual review of yellow flags.

The more useful point is that detection is only half the system. The other half is the interview that follows. Online tests in India are usually a shortlisting step before technical and HR rounds, so a score you cannot reproduce live does not help you get the offer. Preparation on DSA for placements and the aptitude topics placement tests use pays off in both rounds.

How can honest candidates avoid false flags?

Set up your room and computer so the only thing the platform sees is you working. A short checklist:

Risk of false flagPrevention
Face not visible or partially visibleSit facing a light source; keep the camera at eye level
Another person flaggedTest alone in a closed room; tell family the exact time
Looking awayKeep rough paper directly in front of the screen if allowed
Phone detectedKeep the phone out of the room or switched off out of view
Tab switch from notificationsQuit messaging apps, turn on Do Not Disturb, pause updates
Background speechChoose a quiet slot; avoid reading questions aloud

Our guide on what happens if you switch tabs in an online test explains the tab-switch rules in more detail.

Related guides

FAQ

Can online coding tests detect ChatGPT?

They mostly detect the behaviour around using it. Copy-paste into the editor can be blocked or logged, tab switches are counted, Testlify detects browser AI plugins, and HackerEarth's looking-away webcam category targets candidates using a phone or second laptop to reach ChatGPT. Employers also re-test skills in live interviews, where a borrowed answer is hard to explain.

Can a proctored test detect a mobile phone?

Webcam proctoring can. Mercer Mettl lists a mobile phone detected AI flag among seven flags, with screen and camera snapshots every 60 seconds and optional full video recording. HackerEarth's looking-away category was designed for candidates using a phone below the desk. Keep your phone switched off and out of the room during any proctored test.

Does HackerEarth check for plagiarism?

Yes. HackerEarth's March 2024 release notes describe an integrity index that lowers a candidate's score for plagiarism, tab switches and webcam snapshot violations, giving employers one figure to review. HackerEarth can also disable copy-paste in the code editor and block copying of problem statements, depending on the employer's settings.

Will a false webcam flag get me disqualified?

Usually not automatically. Platforms present webcam flags as evidence for the employer to review. Testlify, for example, says minor issues such as background noise produce a yellow flag needing manual review of the recording. Reduce false flags by testing alone in a quiet, well-lit room with the camera at eye level and your phone out of view.

Can online tests detect a second monitor?

Lockdown browsers can. HackerEarth's Smart Browser blocks additional monitors, projectors and TVs, as well as virtual machines and virtual cameras. Mettl Secure Browser will not launch a test while a screen is shared, and connecting a cable to share the screen during the test ends it. Disconnect extra displays before you start.

Final thoughts

Online assessments catch most cheating through a mix of logs, webcam analysis and later interviews, and a flag on an honest candidate is usually reviewed by a person. The reliable way through is preparation, a clean test setup and skills you can show again in the interview. If you want help planning that preparation, try the Asuraa career path planner.

Related articles

Share this article

Continue Reading

Data Science Career Paths

Explore different career trajectories in data science and find your perfect fit.

Read article →

Building Your DS Portfolio

Learn how to create projects that impress hiring managers and showcase your skills.

Read article →

Salary Negotiation Guide

Get the compensation you deserve with our proven negotiation strategies.

Review Your Resume →