Eight in 10 risk managers at large American companies said their organizations experienced employee theft, fraud, or embezzlement in the past 12 months, a new QBE Insurance survey has found, as artificial intelligence reshapes how workplace crimes are both committed and detected.
The survey polled 200 US risk managers at firms with at least $500 million in annual revenue. All respondents had knowledge of employee crimes at their organizations within the past three years.
Billing fraud was the most reported scheme over the three-year window, cited by 36% of respondents, followed by payment and check fraud at 23%, payroll fraud at 19%, and cash or non-cash theft at a combined 23%.
Billing schemes exploit the accounts payable function – the same pipeline through which legitimate vendor payments flow – making them hard to detect.
The risk is heightened when invoices are for services rather than physical goods, as process automation firm MHC Automation has noted, since there is no inventory to cross-check and staff processing hundreds of payments monthly seldom have time to spot anomalies.
The financial toll is well documented. The Association of Certified Fraud Examiners' 2024 Report to the Nations, published in March that year, pegged median losses at $145,000 per case, with the average reaching $1.7 million.
The ACFE estimated a typical organization loses 5% of revenue annually to fraud - implying $25 million in exposure for a $500 million company.
Seniority compounds the damage. ACFE data showed median losses of $500,000 when owners or executives were involved, versus $60,000 for rank-and-file staff.
That pattern is consistent with the QBE survey's finding that more than three-quarters of recent incidents were committed by someone at manager level or higher.
An overwhelming 94% of respondents told QBE they were concerned about employees using AI to commit workplace crimes. Already, 45% of organizations deploy AI-based detection systems, with another 50% planning to follow suit within 12 months.
The threat extends to deepfake-enabled fraud, where AI-generated video and voice impersonation have been used to authorize transactions.
Despite 92% of organizations having tightened internal controls over the past year, nearly half of risk managers said they remain "very concerned" or "extremely concerned" that safeguards are inadequate. The most common measures to curb employee theft were automated monitoring tools, staff training, and internal audits.
On the coverage side, 87% carry crime insurance, with nearly two-thirds planning to increase their limits. QBE pointed to heightened awareness of AI-powered fraud and economic pressures as key drivers.
The Coalition Against Insurance Fraud estimates that fraudulent activity costs Americans up to $308.6 billion annually.
