CRIF launches AI tool to spot tampered onboarding documents

CRIF has launched an AI-powered fraud detection service in the UK to identify tampering in documents used for banks’ and insurers’ customer onboarding. The service inspects IDs, utility bills and bank statements to flag visual edits and inconsistencies in file metadata.

CRIF stated the system combines neural networks, large language models and specialised deepfake-detection models to detect image manipulation, altered text and anomalies in underlying file data that can indicate fraud.

The product targets banks and insurers that process business onboarding. Manipulated documents can be used to conceal poor credit histories, misrepresent a company’s sector or change a firm’s risk profile to obtain loans or more favourable terms. CRIF’s methodology is intended to reveal subtle traces left by both conventional editing software and AI-generated alterations.

Manual document checks remain common in onboarding workflows and can be time-consuming and costly. CRIF estimates manual reviews can account for up to 5% of a bank’s operating costs. The service analyses large volumes of documents quickly and returns a traffic-light risk indicator while keeping human review and final decision-making in place.

CRIF added the product can be embedded in existing onboarding systems and complies with UK regulatory and compliance requirements. The UK launch follows earlier deployments across Europe as part of a wider rollout aimed at helping financial institutions detect evolving fraud techniques without slowing onboarding.

Research CRIF cited alongside the launch found 67% of UK business leaders believe AI-powered services can speed financial decision-making at banks and insurers and could enable more tailored financial products.

Sara Costantini, CRIF’s regional director for the UK and Ireland, commented that fraud is affecting onboarding processes and that AI-based detection is required to identify AI-manipulated documents.