Linguistic Deception Analysis

Text doesn't lie.
The language does.

Candor detects deception in written communications by analyzing linguistic patterns that humans miss. Not keyword matching. Not sentiment scores. Cognitive science applied to every sentence.

Analyze text now ↓

Try Candor live

Paste any text below. Candor will analyze it for deception signals in real-time.

0 / 10,000 characters
🔍

Enter text and click analyze to see deception signals in real-time

What Candor Reads

01

Linguistic Distance

Deceptive text creates measurable distance between the writer and their claims. Candor quantifies pronoun shifts, hedging patterns, and commitment markers that signal evasion.

02

Cognitive Load

Lying is harder than telling the truth. Fabricated narratives show reduced complexity, fewer sensory details, and unnatural temporal structures. Candor catches the strain.

03

Pattern Breaks

People have linguistic fingerprints. When someone shifts from their baseline, whether in formality, specificity, or emotional tone, Candor flags the deviation.

Where deception hides in plain text

Every industry has communications that matter too much to take at face value. Candor reads them for you.

Insurance

Claims Analysis

Score incoming claims for linguistic deception signals before they reach an adjuster. Flag inconsistencies in narrative structure and detail patterns.

Trust & Safety

Content Integrity

Detect fake reviews, fraudulent listings, and manipulative content across marketplace platforms using linguistic pattern analysis.

Legal

Deposition Screening

Analyze testimony transcripts for deception markers. Surface inconsistencies in witness statements that warrant further investigation.

HR & Recruiting

Candidate Screening

Evaluate written responses, cover letters, and assessment answers for authenticity signals and embellishment patterns.

One endpoint. Full analysis.

Integrate Candor into your pipeline with a single API call. No SDK, no dependencies.

POST /api/analyze
curl -X POST https://getcandor.polsia.app/api/analyze \
  -H "Content-Type: application/json" \
  -d '{
    "text": "I honestly have no idea how the damage occurred. The car was parked and I think maybe something fell on it."
  }'

// Response
{
  "success": true,
  "analysis": {
    "score": 62,
    "confidence": 0.78,
    "verdict": "high_risk",
    "signals": {
      "pronoun_distancing": { "score": 55, ... },
      "hedging": { "score": 78, ... },
      "detail_specificity": { "score": 70, ... },
      "cognitive_complexity": { "score": 45, ... },
      "emotional_leakage": { "score": 62, ... }
    },
    "flagged_sentences": [...],
    "summary": "Multiple deception signals detected..."
  }
}
Built on Research

Grounded in cognitive neuroscience, not marketing claims

Candor's models are informed by published research on how language patterns reveal deception in digital communication, including novel ERP findings on truth perception in the brain. This isn't another black-box AI. It's science you can cite.