Industries
The Fraud Scanner combines the Image Trust Layer, Document Trust Layer, and Information Trust Layer — analysing images, documents, and content simultaneously in seconds.
Multi-layer fusion architecture
95%+
Detection accuracy
< 200 ms
Analysis time per scan
20+
Analysis features
~ 4 PT
Integration effort
Image Trust Layer, Document Trust Layer, and Information Trust Layer — each layer brings its specialisation. Together they form the Fraud Scanner: an end-to-end solution that misses no fraud.
Image Trust Layer
AI detection · Manipulation · Heatmaps · Metadata · Content moderation · Damage classification
Document Trust Layer
AI generation · Document forgery · PDF structure · OCR extraction · Classification
Information Trust Layer
OSINT plausibility checks · Image-in-image detection · Contextual validation
Fraud Scanner
Unified product
The Fraud Scanner checks images, documents, and content in parallel in a single scan.
Detection of AI-generated images
ImagesDetection of (AI-)manipulated images
ImagesHeatmap for explainability
ImagesDuplicate check (internal/external)
ImagesReverse image search
ImagesMetadata analysis
ImagesImage-by-image detection
ImagesImage quality analysis
ImagesDetection of AI-generated documents
DocumentsDetection of (AI-)manipulated documents
DocumentsDuplicate check (internal/external)
DocumentsMetadata analysis
DocumentsPDF structural analysis
DocumentsDocument classification
DocumentsPrediction of capture location from pixels
ContentAgentic plausibility checks (OSINT)
ContentOCR data extraction
ContentDetection of minors
ContentDetection of nudity
ContentDetection of QR codes & links
ContentDetection of phone numbers
ContentProcessed in parallel. Completed in seconds.
A single API call or upload triggers all available analyses simultaneously — across all three trust layers. Results arrive within seconds, not minutes.
Benchmarks based on internal tests with over 50M processed media files.
Basierend auf internen Benchmarks mit über 50 Mio. verarbeiteten Medien.
The Fraud Scanner integrates in approx. 4 person-days — as a flexible REST API or as a ready-to-use web interface for manual reviews.
REST API
REST · JSON · Webhook
// Analyse in seconds
const result = await vaarhaft.scan({
file: imageFile,
modes: ["image", "document", "content"],
});
// Response
{
fraudScore: 0.97,
detectedType: "ai_manipulated",
heatmapUrl: "https://...",
layers: LayerResults
}Seamless integration into your existing systems. Comprehensive documentation, SDKs and direct developer support.
Integration typically completed in approx. 4 person-days.
Web Tool
SaaS · No code required
Ready to use immediately after login — no integration required. With heatmap visualisation, PDF reporting, and direct result export.
On-premise available
Deployment in your own infrastructure possible — for maximum data control and compliance. Regular model updates are delivered, and independent fine-tuning of the models is supported.
Fully GDPR compliant. All data is processed on German servers.
Your data is used exclusively for the current analysis and deleted immediately afterwards. No third-party sharing.
We never train our models on your customer data. Your data stays your data.
Deployment in your own infrastructure possible — for maximum data control and compliance. Regular model updates are delivered, and independent fine-tuning of the models is supported.
Our models are continuously updated to counter new AI methods and threat patterns.
Fast implementation thanks to clear API documentation and direct technical support.
From insurance to HR — the Fraud Scanner adapts to your workflow and processes.
Fraud Detection
Fraud Scanner
The Fraud Scanner analyses and detects fraud — giving you certainty whether an image, document, or piece of content has been manipulated or forged.
Fraud Prevention
SafeCam
SafeCam acts even earlier: it prevents manipulated images from being submitted at all — directly at the point of capture.
Together, Fraud Scanner and SafeCam form a seamless anti-fraud chain: detect and prevent.
Learn more about SafeCamThe commonly used image formats JPEG, PNG, TIFF, HEIC and PDF can be processed.
Yes, the service can also analyze several images in a row. Multiple images can be analyzed as a batch when received as a zip file, with individual assessments for each image.
Yes, the Fraud Scanner can analyze images out of PDF files. Furthermore, the Fraud Scanner is also able to analyze document PDF files like invoices, certificates or other text documents.
The images are briefly stored on our servers for analysis and are completely deleted following the analysis.
The results of the Fraud Scanner are not yet court-proof. Even if the detection rates are very high and reliable, the Fraud Scanner is a statistical model that can never guarantee 100% security.
We constantly develop our detection models and adapt them to the latest trends in AI development to ensure reliable detection of new manipulation techniques.
Yes, on-premise deployment is available for organizations with special data protection requirements. Contact us for a tailored setup.
Integration typically takes around 4 person-days. We provide comprehensive API documentation, SDKs, and direct developer support.
Get a detailed insight into how you can protect yourself from image‑based fraud
Kostenlos · Unverbindlich