V
VERDAT
Image authenticity verification API

Detect manipulated and AI-generated images at scale

VERDAT analyzes images for signs of artificial generation, digital manipulation, and metadata inconsistencies. Get probabilistic confidence scores and audit-ready forensic reports.

View sample report
Multi-layer

Forensic analysis combining metadata, visual, and AI detection signals

Explainable

Every score backed by specific detected signals and evidence

Probabilistic

Calibrated confidence scores, not binary verdicts

Audit-ready

Structured reports designed for compliance review

Image fraud is accelerating

AI-generation tools and sophisticated editing software have made it trivial to create convincing fake images. Traditional validation methods cannot keep pace.

Exponential growth in synthetic media

Generative AI tools have democratized image synthesis. Creating photorealistic fake images now requires no technical expertise.

Manual review does not scale

Human analysts cannot keep up with claim volumes. Review backlogs grow while fraudulent submissions slip through.

Legacy tools are insufficient

Metadata checks and basic hash comparisons miss sophisticated manipulations. Modern fraud requires modern detection.

Multi-layer forensic analysis

VERDAT combines multiple detection methodologies to provide comprehensive image authenticity assessment. Evidence-based scoring, not binary verdicts.

Metadata inspection

Analyzes EXIF data, file structure, and encoding patterns for inconsistencies that suggest manipulation or synthetic origin.

Visual forensic signals

Detects compression artifacts, clone regions, lighting inconsistencies, and pixel-level anomalies indicative of editing.

AI-generation detection

Identifies patterns characteristic of diffusion models, GANs, and other generative architectures across major AI image generators.

Probabilistic scoring

Returns calibrated confidence scores rather than binary classifications. Supports risk-based decision thresholds for your workflow.

Structured reports

Generates audit-friendly documentation with detailed findings, supporting evidence, and reproducible analysis methodology.

Explainable results

Every score is backed by specific detected signals. Understand exactly why an image was flagged for further review.

How it works

Integrate image authenticity verification into your existing workflows with a single API call.

1

Submit image

Send an image to the VERDAT API via URL or base64-encoded data. Supports JPEG, PNG, WebP, and HEIC formats.

2

Analyze signals

VERDAT examines metadata, visual artifacts, and AI-generation patterns using versioned detection models.

3

Receive report

Get a structured JSON response with confidence scores, detected anomalies, and supporting evidence for each finding.

response.json
{
  "analysis_id": "ver_a1b2c3d4e5f6",
  "model_version": "2024.12.1",
  "confidence_scores": {
    "ai_generated": 0.12,
    "digitally_edited": 0.87,
    "metadata_anomaly": 0.65
  },
  "detected_signals": [
    {
      "type": "clone_region",
      "confidence": 0.91,
      "location": {"x": 240, "y": 180, "w": 120, "h": 80}
    },
    {
      "type": "exif_inconsistency",
      "confidence": 0.78,
      "details": "Software field indicates editing tool"
    }
  ],
  "audit_report_url": "https://api.verdat.net/reports/ver_a1b2c3d4e5f6"
}

Built for fraud prevention teams

VERDAT integrates into existing risk and compliance workflows across industries where image evidence drives decisions.

Insurance claims

Validate damage photos submitted with property, auto, and liability claims. Detect manipulated evidence before payouts are approved.

  • Pre-screen claim images at submission
  • Flag AI-generated damage evidence
  • Generate audit trails for SIU review

Delivery disputes

Verify photo evidence for missing or damaged package claims. Reduce fraudulent refund requests with automated image analysis.

  • Analyze delivery confirmation photos
  • Detect edited or synthetic complaint images
  • Score dispute legitimacy automatically

E-commerce returns

Validate product condition photos in return and chargeback workflows. Identify patterns of return abuse across customer accounts.

  • Screen return request images
  • Compare against original product photos
  • Integrate with chargeback dispute flows

Built for compliance and audit requirements

VERDAT is designed to meet the documentation, reproducibility, and traceability standards required by legal, risk, and compliance teams.

Versioned detection models

Every analysis includes the specific model version used. Historical results can be referenced and compared as detection capabilities evolve.

Reproducible results

Given the same image and model version, VERDAT returns identical results. Deterministic analysis supports legal and audit proceedings.

Audit-ready documentation

Detailed reports include methodology explanations, detected signals, and confidence calibration data. Ready for regulatory examination.

Data security and privacy

Images are processed in isolated environments and are not retained after analysis unless explicitly requested. Infrastructure designed to support enterprise security requirements.

Developer-first API design

Integrate VERDAT into your existing fraud detection pipeline with minimal engineering effort. Clean REST API with comprehensive documentation.

JSON-based responses

Structured response format designed for programmatic processing and workflow automation.

Designed for low latency

Architecture optimized for fast response times. Async endpoints available for batch processing workloads.

Flexible integration

Works with existing fraud tools, case management systems, and decision engines via REST or webhooks.

Comprehensive documentation

API reference, integration guides, and sample code in Python, Node.js, Go, and Java.

example.py
import verdat

client = verdat.Client(
    api_key="vdt_...",
    base_url="https://api.verdat.net"
)

# Analyze an image
result = client.analyze(
    image_url="https://example.com/claim-photo.jpg",
    include_signals=True
)

# Check confidence scores
if result.scores.digitally_edited > 0.7:
    print("High likelihood of manipulation")
    for signal in result.detected_signals:
        print(f"  - {signal.type}: {signal.confidence}")

# Generate audit report
report = client.generate_report(result.analysis_id)
print(f"Report URL: {report.url}")

Start verifying image authenticity

Join fraud prevention teams using VERDAT to detect manipulated and AI-generated images. Request API access to begin your integration.

No credit card required. Sandbox access included.

Built with v0