LayersRank

Product / Integrity Detection

Trust What You're Evaluating

Remote interviews create opportunities for shortcuts that in-person interviews don't. LayersRank tracks behavioral signals that indicate when responses might not be authentic -- so you can investigate before making decisions.

app.layersrank.com

Integrity Monitor

Session Integrity — Candidate #4829

All Clear

Tab Focus

No switches detected

Response Timing

Consistent with thinking

Copy-Paste

No external paste events

Answer Coherence

Style consistent across Qs

Video Presence

Single person, on-camera

AI Generation

No LLM patterns detected

The integrity challenge in remote hiring

Remote interviews transformed hiring efficiency. A process that took two weeks of scheduling now happens in 48 hours. But remote interviews also created vulnerabilities that didn't exist when everyone was in the same room.

In a face-to-face interview, you know the candidate is answering from their own knowledge. You can see their eyes aren't reading from notes. You can tell no one is whispering answers. The physical presence creates natural integrity.

In a remote interview, those constraints disappear. A candidate can have prepared answers on a second monitor. They can search technical questions. They can paste entire answers from ChatGPT. In extreme cases, someone else can take the interview entirely.

We don't believe most candidates do these things. The vast majority want to earn opportunities based on their actual abilities. But “most” isn't a hiring strategy.

The challenge:

Detecting potential integrity issues without creating a surveillance-heavy experience that treats every candidate like a suspect. The goal is appropriate visibility, not paranoid monitoring. We surface unusual patterns for your review. We don't make accusations or automatic rejections.

What we track

Four categories of behavioral signals, each with clear concern thresholds.

Paste Events

Every paste into a response field is logged: which question, content size, whether it appeared in the final response, and timing relative to question appearance.

Low concern

1-2 pastes, small content, scattered

Moderate concern

3-4 pastes, some on technical questions

High concern

5+ pastes on difficult questions, large blocks, minimal editing

Tab Switches and Window Changes

Logged when candidates navigate away: number of switches, duration, active question, and timing within the question. The analysis considers timing (immediate vs. mid-response), duration (brief vs. extended), and correlation with question difficulty.

Low concern

3-5 brief switches, scattered

Moderate concern

5-8 switches, some extended, some difficulty correlation

High concern

10+ switches, consistently extended, strongly correlated with hard questions

Typing Pattern Analysis

We analyze words per minute, speed consistency, pause patterns, backspace frequency, and time-to-first-keystroke. Authentic composition has a signature: pauses while thinking, bursts as ideas come, corrections, variable speed. Copied content has a different rhythm.

Speed anomalies: 150+ WPM with no paste events suggests undetected copy-paste (avg human: 40-50 WPM)

Consistency anomalies: Robotic same-WPM throughout a 300-word response is unusual

Pause anomalies: Complex questions with no pauses >2 seconds is unexpected

Correction anomalies: Zero backspaces in a long response is statistically unusual

Response Timing

Total time per question, thinking time before first interaction, and comparison to expected duration by difficulty. A comprehensive system design answer in 90 seconds for a difficulty-9 question is a signal worth investigating.

Abnormally fast: Response quality inconsistent with response time

Abnormally slow: Extended delays correlating with tab switches

Inverse difficulty: Easy questions taking longer than hard ones suggests selective lookup

Face verification

Optional

For organizations requiring identity confirmation. Answers one question only: Is the person taking this interview the person who's supposed to be taking it?

Identity confirmation at start

The candidate takes a photo via webcam, compared against a government ID, previous interaction photos, or LinkedIn profile photo. Catches the scenario where someone else takes the interview on the candidate's behalf.

Periodic verification during interview

Optional brief prompts (“Please confirm you're still there”) with a 3-5 second camera capture. Catches the scenario where the legitimate candidate hands off partway through. Configurable frequency.

What face verification does NOT do

No facial expression analysis
No emotion detection
No continuous video recording
No demographic inference

The integrity report

The integrity section summarizes behavioral signals in a clear, actionable format.

Clean Report

INTEGRITY SUMMARY

Paste Events: 2 (minor -- name field, one short phrase)

Tab Switches: 4 (brief, scattered, no pattern)

Typing Pattern: Normal variation

Response Timing: Within expected ranges

Face Verification: Confirmed

FLAG STATUS: NONE

All behavioral signals within normal parameters.

Flagged Report

INTEGRITY SUMMARY

Paste Events: 6

- Q3 (Technical): 287 chars, minimal editing

- Q5 (Technical): 412 chars, no editing

- Q8 (Technical): 194 chars, minor editing

Tab Switches: 14 (avg 38s, 9 within 10s of question)

Typing: Q5 at 127 WPM, Q8 at 143 WPM, 0 backspaces

Timing: Q5 done in 2m14s (expected 6-8m)

FLAG STATUS: REVIEW RECOMMENDED

Concerning patterns on Q3, Q5, Q8.

What flagging means (and doesn't mean)

A flag is information, not a verdict. Behavioral patterns deviated from normal in ways that warrant human review before making advancement decisions.

Possible explanations for flagged behavior

Actual integrity violation

Used external resources, had assistance, or didn't demonstrate authentic capability.

Unusual but legitimate behavior

Exceptionally fast typer. Multi-monitor setup triggering tab detection. Composed in a notes app then pasted.

Technical artifacts

Browser extensions, accessibility tools, or system configurations creating false signals.

Preparation, not cheating

Reviewed similar questions beforehand with practiced responses. Good preparation creates similar patterns.

How to handle flags

1

Review and proceed

Examine patterns, conclude they're not concerning, advance normally.

2

Review and verify

Advance but probe flagged areas live. "Can you walk me through that same problem?" If they reproduce the quality, it was likely a false positive.

3

Request clarification

"We noticed some unusual patterns. Can you help us understand your workflow?" Give honest candidates a chance to explain.

4

Review and decline

Strongly concerning patterns for a high-trust role. Document the specific behavioral patterns that informed the decision.

Privacy-First Design

What we don't do

Integrity detection is designed to surface signals, not to surveil. We explicitly avoid approaches we consider overreaching.

No automatic rejections

Flags go to human reviewers. The platform provides information. Humans make decisions.

No facial expression analysis

Research shows this is unreliable and potentially biased. Face verification is identity-only.

No voice stress analysis

This technology doesn't work reliably and creates legal and ethical concerns.

No audio monitoring

We don't listen for other voices or background sounds. Invasive, unreliable, privacy concerns.

No network traffic analysis

We don't monitor what other sites candidates visit. Only our interface.

No continuous screen recording

We see what happens in our interface. We don't surveil broader computer activity.

No keystroke logging

Typing patterns only within our response fields. No keyloggers, no external monitoring.

No post-session access

The interview ends, our visibility ends. No ongoing device access.

Our philosophy: Detect enough to surface legitimate concerns. Don't invade privacy beyond what's necessary. Treat candidates as professionals unless specific evidence suggests otherwise.

Configuring integrity detection

You control how integrity detection operates for your interviews.

Sensitivity levels

Default

Standard

Flags clearly unusual patterns. Balanced between catching concerns and minimizing false positives. Appropriate for most roles.

Strict

Flags borderline patterns. Higher sensitivity, more false positives. For high-stakes, security-sensitive, or regulated roles.

Minimal

Only extreme anomalies. Lower sensitivity, fewer false positives. When candidate experience is paramount.

Feature toggles

Enable or disable specific tracking per your needs.

Paste detectionOn
Tab switch trackingOn
Typing pattern analysisOn
Response timing analysisOn
Face verificationOff / Start only / Periodic

Alert configuration

Include integrity summary on all reports
Only notify when flags appear
Real-time alerts for extreme anomalies

Candidate disclosure

We strongly recommend transparency. The interview landing page can include a customizable disclosure:

“This interview session tracks behavioral patterns including paste events, tab switches, and typing patterns to ensure evaluation integrity. By proceeding, you acknowledge this monitoring.”

Transparency deters casual cheating, sets expectations, avoids candidates feeling surveilled without knowledge, and addresses legal disclosure requirements in some jurisdictions.

Handling flagged candidates: a framework

01

Review the specific patterns

What exactly was flagged? Paste events on which questions? Tab switches with what timing? Read the details, not just the flag status.

02

Assess severity

Single anomaly or consistent pattern? Concentrated on high-stakes questions or scattered? Does the combination tell a coherent story, or could each signal have an innocent explanation?

03

Consider role context

What's the cost of a false negative (advancing a cheater) vs. a false positive (rejecting an honest candidate)? High-trust roles err toward caution. High-volume roles give more benefit of the doubt.

04

Decide on response

Proceed normally, verify in subsequent round, request re-take under stricter conditions, or decline to advance. Match response to severity and role requirements.

05

Document your reasoning

Whatever you decide, note why. Protects you if decisions are questioned later and helps calibrate your process over time.

Honest Assessment

Does integrity detection work?

An honest assessment of what we know and don't know.

What we can measure

~8% flag rate

Of interviews receive some integrity flag. Of those:

~60% minor flags, typically dismissed after review

~30% moderate flags prompting verification

~10% significant flags influencing decisions

~20% false positive estimate

Based on flagged candidates who were advanced and performed well.

What we believe based on evidence

Patterns we flag correlate with concerning behavior. The combination of paste events, tab switches, timing anomalies, and typing patterns that trigger high-severity flags is unlikely to occur through innocent behavior.

Sophisticated cheating can evade detection. A determined candidate who types pre-memorized answers, uses a separate device, and has invisible help could potentially evade our detection. We catch common patterns, not every possible evasion.

Most candidates are honest. The 92% clean rate confirms this. Actual integrity violations are a small minority -- but that small minority can do real damage if undetected.

The balance: security versus experience

Integrity detection involves real trade-offs. The right balance depends on context.

Too aggressive

Creates a hostile, surveillance-heavy experience. Good candidates who value their dignity may decline. Your employer brand suffers. You lose talent to competitors who treat candidates more respectfully.

Right balance

LayersRank defaults to a balanced approach appropriate for most professional hiring. Adjust sensitivity up or down based on role requirements, candidate relationship, and organizational risk tolerance.

Too permissive

Creates exploitable gaps. Candidates who cheat gain unfair advantage. Evaluations don't reflect actual capability. You make decisions based on false information. Bad hires result.

Frequently asked questions

Do candidates know they're being monitored?

We recommend transparency. Default interview instructions include disclosure of behavioral monitoring. Candidates who proceed have acknowledged the monitoring.

Can tech-savvy candidates bypass detection?

Some evasion is possible. Using a separate device for lookups, typing pre-written answers manually, having someone else in the room whispering answers -- these might not trigger our specific signals. We catch common, easy shortcuts. If common shortcuts are risky, most people who would have cheated casually won't.

What about candidates with disabilities?

Contact us before the interview. We can adjust sensitivity settings, disable specific detection features, or provide alternative formats. Some patterns that trigger flags (unusual typing patterns, extended pauses, use of assistive technology) might be normal for candidates with certain disabilities. We configure appropriately when informed.

Is this legal?

Monitoring candidate behavior during an interview you invited them to take is generally legal, especially with disclosure. However, laws vary by jurisdiction. GDPR in the EU, CCPA in California, and various employment laws may impose requirements around disclosure, consent, and data handling. We recommend consulting your legal team. We provide disclosure templates and data handling documentation to support compliance.

What happens to integrity data?

Integrity data is retained alongside interview responses, following your configured data retention policy (typically 1-2 years). Candidates can request deletion under applicable privacy laws. We honor such requests within required timeframes. Integrity data is not shared externally or used beyond the specific hiring evaluation.

Can I see integrity data for candidates I didn't flag?

Yes. Configure reports to include integrity summary on all candidates, not just flagged ones. This gives you visibility into normal patterns, which helps calibrate your interpretation of flagged patterns.

Hire with confidence

See how integrity detection works in practice. Book a demo and we'll walk through flagged and clean report examples -- exactly what you'd see for real candidates.