Carpe Data: insurance fraud costs rise as frequency falls

New fraud data shows AI-driven detection scaling across 100% of claims as per-incident losses grow and social media evidence dominates.

By Disrupts Editorial Team Tue, 13 October 2026 AI level 2 of 5

AI level 2 of 5: drafted by our AI editorial assistant from the company's announcement; fact-checked, edited and signed off by Mark Walker, Editorial Director. What the levels mean

A bright modern office space features a curved white desk with blue LED strip lighting, supporting three monitors on black articulated arms displaying dark-themed data visualizations, with large windows and blurred glass partition walls in

Carpe Data, a Santa Barbara-based provider of data analytics for the insurance sector, has published its 2025 Online Fraud Insights report, revealing a troubling inversion in the pattern of injury claims fraud across the United States: incidents are becoming less frequent but significantly more costly.

The company's modelled value of contradictory alerts grew from $230.6m in 2024 to $263.8m in 2025, even as referral rates across disability, workers' compensation, and auto lines declined year-on-year. The implication for carriers is that each fraudulent claim that slips through now represents a larger financial exposure than before.

Social media as the primary evidential layer

Carpe Data's analysis, drawn from continuous monitoring of open injury claims across auto, workers' comp, general liability, and disability lines throughout 2024, identifies social media as the dominant source of contradictory evidence. Facebook accounted for 60.5% of contradictory flags, outpacing Instagram at 13.8%, LinkedIn at 7.8%, TikTok at 3.2%, and X at 0.3% combined. Disability claimants showed the highest misrepresentation rate at 8.9%, ahead of workers' compensation at 6.9% and auto at 5.7%.

Geographically, Connecticut, Nevada, South Carolina, Delaware, and Georgia recorded the highest alert concentrations per 100,000 residents, a pattern the report links to the confluence of strong legal networks, population density, tourism, and regulatory arbitrage opportunities. About 36% of fraud referrals surfaced within the first two months after a reported loss, a figure that has remained stable year-on-year.

Tom Rasmussen, VP of Claims Product at Carpe Data, pointed to attorney-driven coaching as a structural driver: "Many individuals committing fraud don't see themselves as criminals, they're often led to believe they're simply maximising their claims. Professional injury attorneys who know how to exploit legal loopholes are coaching claimants into exaggerating their injuries, sometimes without them fully realising the consequences."

The AI detection gap and its cross-sector implications

The report's most strategically significant data point for cross-sector observers is the detection coverage gap. Traditional fraud investigation methods, the report notes, typically analyse only around 5% of open injury claims at any one point in time. AI-driven tools, Carpe Data argues, can extend that coverage to 100% of the open claims portfolio, identifying behavioural inconsistencies at a scale that human review cannot match.

This is not simply an insurance-operations story. The shift from point-in-time to continuous, AI-powered monitoring mirrors a broader pattern visible across financial services, healthcare administration, and public-sector benefits programmes: legacy detection architectures built for sampling are being systematically replaced by always-on machine-learning models trained on behavioural signals. For insurers, the capital case is straightforward, every undetected fraudulent injury claim drains thousands from the book; scaling detection to full-portfolio coverage turns a cost centre into a risk-adjusted asset.

The investment landscape around AI-native insurtech fraud tools is expanding accordingly. Venture and growth-equity capital has been flowing steadily into the claims-intelligence layer of the insurance stack over recent years, with platforms competing on the breadth of open-source signal ingestion (social media, public records, local news) and the speed of alert generation. Carpe Data's findings, particularly the two-month early-alert window, reinforce the commercial thesis that continuous monitoring delivers its greatest marginal value in the earliest phase of a claim's lifecycle, before traditional investigation processes are even triggered.

For macro investors tracking AI adoption curves, the insurance sector represents a relatively late but now accelerating cohort: large, data-rich, and historically under-automated, it is becoming a proving ground for AI applications in complex document and behavioural analysis that have direct read-across to adjacent sectors including legal-tech, public-sector fraud prevention, and healthcare claims adjudication.