Although paper checks are steadily declining as consumers and businesses adopt faster digital payments, check fraud remains a persistent and costly problem.
In 2024, checks were responsible for 30% of all fraud losses, despite making up a far smaller share of transaction volumes. Research from PYMNTS Intelligence and The Clearing House shows that check payments are 31 times more likely to be fraudulent than real-time transactions, highlighting why banks continue to prioritise defences in this area.
Hawk, which offers AML and anti-fraud solutions, recently delved into how AI is transdorming check fraud detection.
Economic pressure has played a significant role in the resurgence. During the COVID period, millions of relief payments were distributed by mail, reinforcing checks as a legitimate source of financial support. For some individuals under financial strain, this association lowered the psychological barrier to committing check-based fraud. Ongoing inflation and cost-of-living pressures have kept this stress firmly in place, sustaining fraud activity even as employment levels recovered, Hawk explained.
Fraudsters are also benefiting from unprecedented access to personal data. Social media platforms encourage users to share information such as birthdays, employers and locations, creating valuable intelligence for criminals. Beyond this, the dark web has evolved into a large-scale marketplace for stolen data. In 2024 alone, 1.9 million stolen US bank checks appeared on underground platforms, providing fuel for increasingly organised fraud operations, it said.
Generative AI has further accelerated the threat. With stolen bank details and a handful of check images, criminals can now create convincing counterfeit or altered checks using widely available GenAI tools and basic printing equipment. These are often deposited through mobile remote deposit capture, allowing fraudsters to move quickly before traditional controls detect anomalies.
Despite digital alternatives, checks remain entrenched in B2B payments, Hawk said. Businesses continue to view them as familiar, traceable and easy to reconcile. In 2024, 33% of B2B payments in the US were still made by check, representing a significant volume of paper-based transactions exposed to fraud risk, it said.
AI has already transformed how financial institutions respond. Computer vision and image forensics analyse visual elements such as handwriting, fonts and layout consistency, while statistical models detect subtle structural anomalies. When combined with consortium data shared securely across banks, these systems prevent fraudsters from reusing the same check templates at multiple institutions.
Cross-channel analysis strengthens this further by linking check activity with ACH, wire and card transactions. Rather than assessing each payment in isolation, AI builds a holistic view of account behaviour, flagging coordinated patterns that indicate fraud. Behavioural biometrics add another layer, identifying whether a mobile user behaves like the genuine account holder during a deposit attempt.
Looking ahead, the focus is shifting towards self-service model development, insider fraud detection and more predictive risk scoring. Platforms that allow institutions to train AI models on their own data will be better equipped to adapt as fraud evolves. Hawk’s partnership with Mitek reflects this direction, combining advanced image forensics, consortium intelligence and AI-driven anomaly detection tailored to each institution’s unique risk profile.
For more insights, read the full story here.
Read the daily FinTech news
Copyright © 2026 FinTech Global
