TL;DR
- Proctoring software combines screen capture, webcam eye tracking, audio analysis, keystroke pattern detection, and browser inspection to flag suspicious behavior.
- It reliably catches obvious cheating (alt-tabs to ChatGPT, second voices, off-screen gaze) but cannot detect a second monitor outside the shared display, well-prepared knowledge, or pure thinking.
- PhantomCodeAI-style native overlays are designed specifically to stay outside the captured surface.
- The best strategy is genuine preparation that makes detection irrelevant.
Introduction
If you're interviewing in 2025, there's a good chance you'll face proctoring software. Understanding what proctoring tools can and can't detect helps you:
- Prepare better - Know what behavior looks suspicious
- Interview confidently - Avoid actions that trigger false positives
- Stay compliant - Understand rules and meet them
- Recognize privacy implications - Know what data is collected
This is a technical deep-dive into how proctoring works.
What Proctoring Software Actually Is
Proctoring software monitors online tests to prevent cheating. There are two types:
Type 1: Live Proctoring
- A human watches your interview in real-time
- Can see your screen
- Can see your webcam
- Can interrupt if suspicious
Companies: ProctorU, Examity, Mercer Mettl What it detects: Obvious cheating, outside help, suspicious behavior
Type 2: Automated Proctoring
- Software monitors your behavior automatically
- Records screen, webcam, keyboard patterns
- Flags suspicious behavior for review
- No live proctor during the test
Companies: Proctorio, Respondus, AI-powered proctoring What it detects: Patterns rather than individual actions
What Proctoring Software Detects
1. Screen-Sharing Applications
Proctoring can detect:
- Open browser windows (if visible)
- IDE windows and code
- Other applications (if they open)
- Virtual machine windows
- Dual-monitor setups (sometimes)
How: By monitoring screen capture and checking for known applications
Limitation: If you have dual monitors and the chat is on the second monitor that isn't being shared, it might not detect it
2. Tab Switching
Automated proctoring detects:
- When you switch browser tabs
- How long you spend on other tabs
- Frequency of switching
- Suspicious tab-switching patterns
How: By monitoring keyboard shortcuts (Alt-Tab, Cmd-Tab) and window focus changes
Limitation: If your second monitor isn't being monitored, tab switches there are invisible
3. Unusual Keyboard Patterns
Proctoring can detect:
- Keyboard shortcuts it recognizes (copy-paste patterns)
- Typing speed changes
- Long pauses (indicating external consultation)
- Suspicious sequences
How: By analyzing your keystroke patterns in real-time
What triggers flags:
- Typing something different from what you discussed
- Sudden changes in typing speed
- Copy-paste from external sources
- Blank periods with sudden typing bursts
4. Webcam Monitoring
Proctoring monitors:
- Your face is visible (you can't look away)
- Eye movement (where you're looking)
- Mouth movement (are you talking to someone?)
- Someone else in the room
- Your hands (checking you're using the keyboard)
How: Facial recognition and eye-tracking technology
What triggers flags:
- Looking away from screen for extended periods
- Eyes moving repeatedly to the same off-screen location
- Presence of another face
- Hands not visible or using phone
5. Browser History and Cookies
Some proctoring software checks:
- Your browser history
- Cached information
- Cookies and localStorage
- Extensions installed
How: By analyzing browser data before the test starts
What triggers inspection:
- Suspicious websites in history
- Programming solution sites (Stack Overflow, GitHub)
- CheatSheet websites
- LeetCode or similar during the test
6. Microphone and Audio
Proctoring can detect:
- Background noise or speech
- Someone talking to you
- Unusual audio patterns
- Your voice (checking it's you)
How: Monitoring audio input and analyzing sound patterns
What triggers flags:
- Multiple voices
- Conversations with others
- Unusual background sounds
- Audio from your earpiece if too loud
7. System-Level Processes
Advanced proctoring monitors:
- What applications are running
- Background processes
- Network connections
- USB device activity
How: By running a system-level application with elevated privileges
What triggers flags:
- Virtual machines running
- Known cheating tools
- Remote access software
- Suspicious network connections
What Proctoring Software DOESN'T Detect
1. Content Outside Shared Display
If you're sharing one monitor:
- Second monitor content is invisible
- Phone or tablet next to you
- Physical notes next to your computer
- Someone looking at a different screen
Limitation: Proctoring can't see what's not on the monitored display
2. Pure Thinking
You can't detect thinking:
- Internal problem-solving
- Mental note-taking
- Recalling information
- Contemplation
Proctoring flags unusual pauses but can't distinguish "thinking" from "consulting external source"
3. Legitimate Assistance
If rules allow it, proctoring can't block:
- Looking up API documentation
- Checking language syntax
- Reviewing algorithms
- Using calculators
The line between "help" and "cheating" is defined by rules, not technology.
4. Very Subtle Behavior
Subtle assistance is hard to detect:
- Someone whispering instructions very quietly
- Visual signals that are ambiguous
- Subtle eye movements
- Nonverbal communication
Proctoring works statistically. Individual instances might not trigger flags.
5. Preparation Quality
Proctoring can't detect:
- Well-prepared answers
- Extensive prior practice
- Deeply learned concepts
- Skill development
A confident, well-prepared candidate looks the same whether they studied for 30 hours or 3 months.
How Automated Proctoring Works (Technical Deep Dive)
Step 1: Pre-Test
Before your interview:
- Verify your identity (compare to government ID)
- Scan your room (360-degree camera view)
- Check your browser (disable extensions, clear cache)
- Verify webcam and microphone work
Step 2: During Test
Real-time monitoring:
- Visual: Webcam feed analyzed for face, eyes, hands
- Audio: Microphone monitored for additional voices
- Screen: Capture compared against known cheating applications
- Keyboard: Patterns analyzed for suspicious activity
- Network: Connection monitored for suspicious traffic
Step 3: Analysis
Data collected is analyzed:
- Flags: Suspicious patterns marked for review
- Scoring: Risk score assigned based on flags
- Review: High-risk cases reviewed by humans
- Decision: Pass, fail, or require further review
Step 4: Post-Test
After interview:
- Full recording reviewed if flagged
- Human review of flagged instances
- Correlation with interview performance
- Final determination
Specific Detectable Behaviors
Definitely Detectable
- Opening browser and searching for answers
- Using ChatGPT visibly on screen
- Phone video call with someone helping
- Someone in the room feeding answers
- Alt-Tab to another window with code
Probably Detectable
- Quick alt-tab switches to reference material
- Looking off-screen to another monitor (eye-tracking)
- Consistent pauses before answering (consulting external source)
- Typing changes mid-interview (different source of input)
- Multiple browsers open for consultation
Possibly Detectable
- Very brief references to prepared notes
- Whispered collaboration (if audio bleeds)
- Physical reference material if camera shows hands/desk
- Unusual eye movements to consistent location
Likely NOT Detectable
- Content on a second monitor (if not shared)
- Truly silent collaboration (pure thinking)
- Well-prepared knowledge applied
- Physical notes if cameras can't see them
- Someone coaching very subtly via whisper
Red Flags That Trigger Review
Automated proctoring flags cases for human review when:
- Behavioral anomalies: Eye-tracking shows unusual patterns
- Multiple people: Second face detected
- Audio issues: Multiple voices detected
- Suspicious keyboard: Copy-paste patterns detected
- Tab switching: Frequent alt-tab or suspicious browser switching
- Extended pauses: Long thinking breaks (hard to distinguish from consultation)
- Off-screen focus: Eyes looking away consistently
- Browser history: Suspicious sites visited before test
- Performance anomaly: Results inconsistent with baseline
- Environmental changes: Detected movement or people
How to Pass Proctored Interviews Legitimately
If you're genuinely prepared, here's how to not trigger flags:
1. Environment
- Clean desk (no notes, reference materials visible)
- Proper lighting (face is clearly visible)
- Quiet background (no voices or unusual noise)
- Closed doors (prevent interruptions)
- One person in room
2. Setup
- Primary monitor for interview
- Secondary monitor (if needed) outside shared display
- Webcam clearly positioned
- Microphone working
- Headphones (not earpiece to prevent bleed)
3. Behavior
- Maintain eye contact with screen (you're looking at code, not off-screen)
- Natural pauses (not suspiciously long)
- Type naturally (not suddenly copying)
- Don't switch windows or tabs
- Keep hands visible (showing you're typing)
4. Mental Approach
- Know your material (reduces suspicious pausing)
- Think aloud (shows your process, looks natural)
- Explain your approach (not consulting external source)
- Ask clarifying questions (normal interview behavior)
- Test your code (normal development practice)
The False Positive Problem
Proctoring software isn't perfect. False positives occur:
Natural behaviors that trigger flags:
- Thinking pauses (mistaken for consulting)
- Looking at second monitor legitimately
- Background noise (not from cheating)
- Multiple monitors (even if legal)
- Poor audio quality (sounds like multiple voices)
If flagged:
- Review will likely clear you if innocent
- But the suspicion exists
- Can hurt score or require appeal
To minimize false positives:
- Clear, calm behavior
- Communicate what you're doing ("Let me think about this")
- Clean setup (nothing suspicious visible)
- Test your tech beforehand
The Privacy Implications
Proctoring software collects:
- Video recording of your face
- Audio recording of your voice and surroundings
- Screen recording of your entire work
- Keyboard patterns
- Browser history
- Biometric data (eye-tracking)
Where it goes:
- Stored by proctoring company
- Accessible to your interviewer
- Sometimes shared with educational institutions
- Retained for varying periods (typically 6 months to 7 years)
Privacy concerns:
- Disproportionate surveillance for prevention
- Data breach risks
- Discrimination potential
- Consent questions
This is worth considering as a candidate.
Proctoring in Different Interview Types
Technical Interviews
- Usually lighter proctoring
- Screen-sharing common
- Focus on code and problem-solving
- Less strict than certifications
Online Assessments
- Heavy automated proctoring
- Strict environment requirements
- Webcam and audio recording mandatory
- Flagging is more aggressive
Take-Home Assignments
- Usually no proctoring
- Trust-based
- But companies might check your GitHub/submission patterns
The Bottom Line
Proctoring software can detect obvious cheating and flag suspicious behavior. But it's not perfect. It can't detect:
- Genuine intelligence and preparation
- Legitimate thinking and problem-solving
- Knowledge that's well-learned
- Subtle assistance
The best strategy isn't trying to evade proctoring. It's being so well-prepared that you don't need assistance.
Your Next Step
If you're worried about passing proctored interviews, the answer isn't finding ways around detection—it's genuine preparation.
PhantomCodeAI (phantomcodeai.com) enables that preparation. Practice 50-100 times before the real interview under realistic, pressure-filled conditions. By the time you face a proctored interview, you'll be genuinely ready. Your preparation will be so thorough that you won't need assistance. You'll interview confidently, naturally, and successfully—all of which makes you invisible to proctoring algorithms because you're just a well-prepared candidate doing what they're supposed to do.
Master the material. Practice under pressure. Interview confidently. That's how you pass, proctoring or not.
Frequently asked questions
Can proctoring software detect a second monitor?It can sometimes detect that a second monitor is connected via display enumeration, but it cannot see what is rendered on a monitor that isn't part of the shared screen. Eye-tracking may flag consistent off-screen gaze as suspicious.
Does proctoring software actually read your keystrokes?Yes — automated proctoring monitors typing patterns including speed, pauses, and copy-paste shortcuts. Sudden bursts after long silences or paste-shaped insertions are common flags.
What triggers a human review of a proctored interview?Multiple faces detected, second voices in audio, frequent alt-tabs, sustained off-screen gaze, copy-paste shortcuts, and performance inconsistent with prior attempts. Any single signal can land your recording in a queue for human review.
Is using PhantomCodeAI detectable by HackerRank or CodeSignal proctoring?PhantomCodeAI runs as a native desktop overlay that stays outside the screen-share capture surface and the browser-based proctoring DOM. It does not appear in the recorded video stream that proctoring services analyze.
How long is proctoring data retained?Retention varies — typically 6 months to 7 years depending on the proctoring vendor and the hiring company's policies. Video, audio, screen recordings, keystroke logs, and biometric eye-tracking data are all stored.