Around 12% of loan applications processed through Ocrolus contain suspicious activity, according to the US-based AI workflow and analytics provider for lenders, which has expanded the fraud detection capabilities of Ocrolus Detect. The update integrates Resistant AI's Documents forensic engine with Ocrolus' existing analysis of income documents, adding a further layer of inspection for bank statements, pay stubs, W-2s, and other financial records. The expanded tool is now live across Ocrolus' customer base and serves lenders, financial institutions, and fintech and consumer platforms.
Forensic checks target AI-generated documents
The integration responds to the spread of AI tools that make it quicker and easier to produce convincing fraudulent financial records. Resistant AI's engine, trained on threat patterns observed across more than 200 million documents worldwide, can identify deepfakes and files created with generative AI that are difficult to spot through visual review or traditional checks. It pairs algorithmic signals drawn from document content with metadata and structural analysis, allowing lenders to assess both what a document says and how it was produced. Resistant AI said the wide availability of AI-powered forgery tools, combined with recurring personal data leaks, means document integrity checks are now needed alongside content verification.
Detect also supports batch processing of multiple documents and offers simplified outcome categories, such as high, medium, or low risk, mapped to recommended actions. Customers can continue using its existing Authenticity Score, while each result is explained through structured flags and reason codes and can be configured to fit a client's review process and risk tolerance. According to Ocrolus, the outputs are designed for fraud analysts who need evidence and visual highlights, underwriters who require clear verdicts, and risk and policy teams focused on precision, recall, and governance.
Fraud losses weigh on lenders
Fraud can account for up to 15% of annual losses for small and medium-sized business (SMB) lenders, according to the American Bankers Association, meaning even a modest reduction in fraudulent applications funded can affect a lender's results. Ocrolus said its customers are under pressure to make faster decisions while dealing with more advanced document manipulation. Beacon Funding Corporation, an equipment finance provider using Detect, said fraud in its sector has outpaced what manual processes can handle, and that the tool allows it to identify fraud at scale without slowing down decisions for clients.
