.png)
Bundling image forensic analysis
In the Fraud Scanner in particular, we combine a variety of complex image forensic analysis to detect different forms of fraud in connection with damage images.
Efficient cost cutting
By more specifically controlling suspicious damage patterns and reducing the effort required for manual image checks, we reduce your costs significantly.
Reduction of fraud rate
Prevent more fraud attempts thanks to new solutions for detecting and preventing manipulated or AI-generated damage patterns.
Uncomplicated Integration
Whether as a support tool for claims settlement on the table or as an optimization for fully automated dark processing, our solutions can be easily and individually integrated into your claims settlement processes.
Automatically. Fast. Reliable.
Detect image-based fraud immediately and save costs in claims settlement
Optimize your input management by using the Fraud Scanner to check the credibility of all damage images.
FRAUD SCANNER FEATURES
Detect dubious damage patterns automatically and comprehensibly
Detection of AI-generated images
Prevent more fraud attempts thanks to new solutions for detecting and preventing manipulated or AI-generated damage patterns.
The Fraud Scanner detects images that were artificially generated by image generators and show content that never existed in reality.
Detection of (AI) edited images
Images that have been subsequently edited by AI or other software are recognized with this feature.
The Fraud Scanner detects images that have been subsequently edited by AI or other image editing software and therefore no longer represent the completely credible scenery of an image. Only edits that distort the content are detected, i.e. those in which something dubious is added to the image or something is retouched. However, the Fraud Scanner does not respond to changes in contrast or brightness.
Marking of image edits
With this feature, the conspicuous and manipulated areas in the image are localized and highlighted in color.
The color markings allow you to quickly and at a glance see where edits have been made in the image and, based on this assessment, initiate the following steps.
Metadata analysis
The metadata for each image is extracted and displayed in an orderly manner so that you can quickly obtain all important additional information about the image.
This feature extracts and displays metadata such as location, time, device or image format. Remember that metadata can easily be changed later and therefore does not provide a reliable basis for drawing conclusions about the credibility of the image. Unfortunately, changes to the metadata cannot be tracked.
Duplicate check
The Fraud Scanner uses this similarity comparison to check whether the analyzed image has already been submitted to you or another insurance company in our database.
Without saving the image ourselves, we will check for you whether the image has already been submitted once. Slight changes in viewing angle have no influence on successful recognition.
Internet reverse search
This feature checks whether the submitted image comes from the internet and has already been uploaded there once.
The Fraud Scanner detects reused images from the internet and even shows you the page where the image was found via a link.
Choose your type of Fraud Scanner integration
API
Integrate the Fraud Scanner where it creates the most added value in your individual processes.
With little effort and our detailed API documentation, you can easily integrate the software interface into your systems and control the call from there.
Low integration effort
Adaptable to your existing systems and processes
Combined use with other fraud detection methods possible
Analyze Cockpit
Get quick and easy access to all features in one place.
With just a few clicks you can access all the features of the Fraud Scanner in our user interface and check images for credibility.
Quick image analysis via drag-and-drop
Clear visualization of the analysis results
Creation of a regular monthly report
Frequently asked questions about the product
-
Which image formats can be processed?The commonly used image formats JPEG, PNG and TIFF can be processed.
-
Can image sequences also be processed, for example in a zip. file?Yes, we can also analyze several images in a row. If we receive the images from you as a zip. file, we can process them and give you the assessment for each image individually.
-
Are the results of the Fraud Scanner court-proof?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.
-
Can PDFs also be processed?Processing images in a PDF document is a further feature currently being developed and is expected to be available in March 2024.
-
Which AIs are recognized?Our service for you is that we constantly develop our decision models and constantly adapt them to the latest trends in AI development. We therefore stay up-to-date with AI development for you. We focus primarily on publicly accessible AIs with the ability to create convincing photorealistic deepfakes.
-
Are the images stored at VAARHAFT during or after the analysis?The images are briefly stored on our servers for analysis and are completely deleted following the analysis. Thanks to our proprietary photo hash algorithm, the images do not have to be stored on our servers for the duplicate check.