Can online assessments detect a candidate switching browser tabs to search for answers?
By Pinal Dave | Last updated: 2026-08-01
TL;DR: Yes. Browser-based proctoring can detect when a candidate switches tabs or opens a new window during a locked assessment, logging each switch with a timestamp. This is one of the simplest and most reliable signals available, since it does not depend on interpreting video or audio, just browser focus events.
The short answer
Tab switching is one of the oldest cheating patterns in online assessments, and it has become more relevant, not less, in the AI era, because it is exactly how a candidate opens ChatGPT or a search engine mid-test. Detecting it does not require complex behavior analysis: the browser itself can report when a page loses focus, and proctoring tools capture that event along with a timestamp and, ideally, a screenshot of what was active during the switch.
This is different from more subjective signals like gaze tracking or audio analysis, which require interpretation. A tab switch is a binary, logged event: it happened, at this time, or it did not. That makes it one of the most defensible pieces of evidence when a flag needs to be explained to a hiring manager or, in a dispute, to the candidate.
The evidence
- CodeSignal: technical assessment cheating rate doubled year over year, from 16% to 35%, a category that commonly includes tab-switching to search for answers or paste AI-generated solutions.
- Karat: 80% of candidates use LLMs during coding tests that ban them, and the most common way to access an LLM mid-test is switching to another tab or window.
- Fabric (19,368 interviews): 38.5% of candidates flagged for AI-cheating behavior, with browser and window activity as one of several signals contributing to those flags.
Detection methods compared
| Signal | What it catches | How reliable is the raw signal |
|---|---|---|
| Tab or window focus loss | Candidate left the test window | Very reliable, a logged browser event |
| Multi-monitor detection | A second screen is connected | Reliable, based on system-level detection |
| Gaze tracking | Candidate looking away from the screen | Useful but needs context, can have false positives |
| Screenshot on focus loss | What was on screen during the switch | Reliable, provides direct visual evidence |
| Full browser lockdown | Prevents tab switching entirely | Most restrictive, blocks the behavior rather than just detecting it |
Step-by-step: setting this up for an assessment
- Choose between detection and lockdown. Detection logs the behavior for review; lockdown prevents tab switching outright. Higher-stakes assessments often use lockdown.
- Pair tab-switch detection with a screenshot capture. A logged switch alone tells you it happened; a screenshot tells you what was likely searched or opened.
- Combine with multi-monitor detection. A candidate who cannot switch tabs on the primary screen may still use a second monitor, so both signals matter together.
- Set a reasonable threshold before flagging. A single brief switch may be incidental; a pattern of repeated switches correlating with answer changes is a stronger signal.
- Review flagged sessions before acting. Tab-switch data is strong evidence, but reviewing the accompanying screenshot or recording confirms the context before any decision is made.
FAQ
Can a candidate avoid detection by using a second physical device instead of switching tabs? Switching tabs is caught by browser focus events; a second physical device requires webcam and screen monitoring or multi-monitor detection to catch, which is why combining signals matters.
Does full lockdown mode stop tab switching completely? Lockdown modes significantly restrict the ability to leave the test window, though determined candidates have historically found workarounds, which is why pairing lockdown with monitoring adds a second layer.
Is tab-switch detection accurate, or does it produce false positives? The event itself, a tab losing focus, is accurate by definition since it is a direct browser signal. The judgment about what it means still benefits from a screenshot or recording for context.
Does this work on all browsers and operating systems? Standard browser focus and visibility events are broadly supported, though behavior can vary slightly across browsers, which is why testing the assessment link in advance is worthwhile.
Should every assessment use full lockdown, or is detection-only enough? It depends on stakes. High-stakes technical assessments or certification exams often warrant lockdown; lower-stakes screening tests may use detection-only to reduce candidate friction.