Socure, the identity-verification firm that counts 18 of the top 20 US banks among its clients, will publish its inaugural Fraud Insights report on 19 October. The study draws on data from Socure’s consortium network, which covers more than 20 billion transactions, 620 million accounts and 178 million unique identities, representing roughly 72% of the adult US banked population. That dataset is paired with a global survey of 1,253 consumers conducted through research platform Prolific.
The headline conclusion is structural rather than incremental: AI is not simply making existing fraudsters more productive. It is recruiting people who would not otherwise have committed fraud at all.
The AI-Fluent Fraudster
Socure’s report introduces a distinct profile it labels the “AI-Fluent Fraudster.” More than half of individuals in this cohort show no evidence of fraudulent behaviour in the absence of AI tools, and their baseline fraud rate, at 4.1%, is statistically close to that of the general population at 5.0%. The implication for fraud and identity teams is significant: traditional behavioural signals built around repeat offenders and known fraud patterns may not flag this group before the first incident.
The scale effect compounds the problem. AI-Fluent Fraudsters committed roughly three times as many fraudulent acts as traditional offenders, and when breadth and frequency are combined, total fraud volume was approximately six times higher. Some 50.6% of AI-Fluent Fraudsters committed at least one fraudulent act ten or more times, a repeat-offence rate 5.5 times that of traditional fraudsters.
The motivational profile is also distinct. Some 41.8% cited curiosity as their primary reason for committing fraud, wanting to test what AI could do rather than acting from financial desperation, a rate nearly double that of traditional fraudsters. The report draws a parallel with Napster: just as downloading music via peer-to-peer networks felt psychologically removed from shoplifting a CD, 15.8% of AI-Fluent Fraudsters said using AI to generate fake hardship letters or falsified bank statements did not feel like they were personally responsible for the act. That figure is close to three times the equivalent rate among traditional fraudsters.
Gen Z is the fastest-growing segment within this cohort. The typical AI-Fluent Fraudster, according to the data, is male and under 30.
Regulatory and industry read-across
The findings arrive at a moment when fraud liability frameworks are tightening on both sides of the Atlantic. In the UK, reimbursement rules for authorised push payment fraud shifted material liability toward payment firms in late 2024, raising the commercial stakes of accurate detection. The EU’s revised Payment Services Directive, PSD3, extends similar logic across member states. Meanwhile the US Consumer Financial Protection Bureau has faced sustained pressure to clarify liability standards for first-party fraud, a category this report directly addresses.
For compliance and fraud-operations teams, the Socure data sharpens a difficult question. If the new cohort of AI-enabled fraudsters is behaviourally indistinguishable from legitimate customers at the point of onboarding, the verification and monitoring controls designed around known fraud rings and synthetic identity networks are, at best, insufficient. Identity verification providers, fraud-tooling vendors and the banks that rely on both will need to revisit model training sets that may not yet reflect the AI-Fluent profile at scale.
The post Socure Report: AI is Creating New Fraudsters, Not Just Better Ones appeared first on The Fintech Times.
