Monday, November 9th, 2026 — Insurance companies are generating more fraud investigation leads through automated detection systems, but their special investigation units (SIUs) are seeing little growth in staffing. The growing reliance on technology has also exposed a substantial difference between the volume of automated alerts and the number investigators consider worth pursuing.
According to the Coalition Against Insurance Fraud's 2024 Insurer SIU Benchmarking Study, conducted with Aon across 35 property and casualty carriers, automated detection tools accounted for 45% of all SIU referrals, an increase of six percentage points from 2022. During that period, average SIU staffing increased by just 0.5%.
The study found that only 36% of automated referrals were accepted for investigation, compared with 70% of referrals submitted by claims adjusters. Although automated tools accounted for nearly half of incoming referrals, they represented only 29% of accepted cases.
Investigative workloads have also increased. The average number of accepted referrals per investigator rose from 162 annually in the 2022 study to 174 in 2024.
The findings have prompted insurance fraud specialists to recommend changes in how carriers evaluate suspicious claims and distribute investigative work.
Brent Walker, director of government relations for the Coalition Against Insurance Fraud, recommends assigning different levels of scrutiny according to the severity and complexity of an alert. Routine concerns could remain with claims adjusters or underwriters, while more substantial allegations would move to desk investigators, field investigators or specialized major case units.
The approach would allow insurers to reserve experienced investigators for cases that require extensive evidence gathering, interviews or analysis of potentially coordinated fraud schemes.
Carriers are also expanding their detection capabilities beyond traditional rules-based systems. Celeste Dodson, the Coalition's director of member relations and initiatives, described the increasing use of predictive analytics, image analysis and network mapping to identify suspicious activity.
Network analysis and entity resolution tools can reveal relationships among individuals, businesses and transactions that might otherwise go unnoticed when claims are examined separately. These technologies are particularly useful when investigating organized fraud, synthetic identities and account takeovers.
The Coalition also warned that generative artificial intelligence has made certain fraudulent activities easier to carry out. Michelle Rafeld, the organization's executive director, said criminals no longer need the same technical expertise previously required to develop sophisticated fraud schemes.
That concern extends beyond claims departments. Fraudulent activity can originate through underwriting, call centers, payment processing and compromised customer information. Coalition leaders recommend stronger coordination among these departments to help investigators identify connected activity.
Insurers must also consider regulatory requirements when deploying AI-based detection systems. Coalition representatives cautioned that outside technology vendors must comply with applicable state insurance laws, including restrictions on data usage and requirements for human oversight.
The organization supports combining existing carrier systems with specialized outside technology rather than relying exclusively on either internally developed software or commercial products.
For claims professionals, the growing use of automated fraud detection means that identifying suspicious activity increasingly involves both traditional investigative judgment and technology-generated evidence. The Coalition recommends feeding investigative outcomes back into detection models so insurers can improve referral accuracy, monitor potential bias and reduce unnecessary investigative work.
