Clyravision: An entire forensics team in one system

Every pixel tells a story. However, in the age of generative AI, not all of them tell the truth. Today, it is no longer enough to recognize a manipulated image. The greater challenge lies in understanding how exactly an image has been altered, analyzing the context and providing clear evidence for a conclusion.

Over the past six months, Merantix Momentum has developed Clyravision - a visual forensics system designed to meet this challenge. Unlike most deepfake detectors, which only provide a single score, Clyravision works like an experienced human analyst. It examines an image from different angles, combines the results and explains precisely how an assessment was made.

Traditional recognition systems work in a single pass: upload the image, analyze it for a metric and generate the score. However, this approach does not capture the complexity of manipulations, which can range from AI-generated faces to real photos taken out of context to classically edited images.

Clyravision takes a different approach. It relies on a multi-agent architecture in which specialized detectors for forensic analysis, semantic consistency and metadata analysis work in parallel. Each tool provides clues that are combined into a coherent chain of evidence to determine whether an image is authentic, manipulated or misused.

This approach is based on three basic principles: Transparency, traceability and modularity. Even with the first prototypes, the aim was to make the decision-making process visible to the user. Instead of an opaque score, Clyravision provides visual highlighting, source comparisons and tool-specific confidence intervals so that users can clearly understand the evidence behind each judgment.

The possible areas of application are diverse. In the media, Clyravision can verify user-generated content before misinformation spreads. In the financial sector, it can uncover manipulated documents and fraudulent claims. Authorities and law enforcement agencies can use it to validate the authenticity of visual evidence for investigations and court proceedings. It can also ensure image authenticity in the insurance industry - for example, to validate accident photos, proof of loss or medical records to ensure they have not been tampered with prior to use in claims, valuations or for regulatory compliance.

Clyravision will soon be available to a wider audience. Organizations and individuals in need of reliable and traceable analysis of images can contact us to learn how our system can support you in your work. Follow us to stay up to date on new features, use cases and our upcoming awareness campaign.

‍