Deepfake fraud climbs for 44% of anti-fraud teams: report

Just 7% rate their organizations more than moderately prepared for AI-powered fraud

Deepfake fraud climbs for 44% of anti-fraud teams: report

Deepfake social engineering was the AI-powered scheme most often reported as having increased significantly over the past two years, cited by 44 percent of anti-fraud professionals, followed by consumer fraud and scams at 38 percent.  

The Association of Certified Fraud Examiners (ACFE) found in its 2026 Anti-Fraud Technology Benchmarking Report that 7 percent of the respondents believe their organizations are more than moderately prepared to detect or prevent AI-powered fraud. 

Banking and financial services accounted for 23 percent of respondents’ organizations and the United States and Canada for 43 percent, the largest region in the sample. 

Looking ahead two years, 55 percent expect generative AI document fraud or forgery to increase significantly, 55 percent say the same of deepfake social engineering and 54 percent of deepfake digital injection. 

John Warren, president and chief executive officer of the ACFE, said in a SAS announcement marking International Fraud Awareness Week that fraudsters hold “a speed, scale and sophistication advantage that’s challenging yesterday’s defenses.”  

As per Warren, fighting these threats takes “sharper detection tools,” investment in the people who use them and “an informed public.” 

Twenty-five percent of organizations use AI or machine learning in their anti-fraud data analysis, up from 18 percent in 2024. 

A further 28 percent expect to adopt it within one to two years.  

Generative AI is in use at 16 percent, with 58 percent planning future use. 

Eighty-two percent rated explainability or auditability important when adopting generative AI, while 6 percent felt completely confident explaining how AI or machine learning models reach anti-fraud decisions.  

Seventy-five percent rated bias or lack of fairness important, and 18 percent said their organizations test models for bias. 

Stu Bradley, senior vice president of risk, fraud and compliance solutions at SAS, said there is “a growing gap between using AI and governing it.”  

As per Bradley, fairness, explainability and governance “must mature alongside fraud detection capabilities or innovation stalls.” 

Budget or financial restrictions were a major or moderate challenge for 83 percent, the most common constraint on implementing new anti-fraud technology.  

Fifty-five percent expect their anti-fraud technology budget to rise over the next two years, 17 percent of them significantly. 

FRAML research by Datos Insights and SAS found 86 percent of financial institutions report missed links between fraud and money laundering, with 71 percent converging detection technology and 13 percent aligning governance, policies and organizational structure. 

An Asia Pacific insurer facing a surge in fraudulent receipt images worked with SAS on a synthetic image detection tool combining computer vision, text extraction and AI reasoning, which accepts images that pass screening, rejects identified fakes and routes uncertain cases to human review.  

In February, Australia’s largest bank referred about A$1bn in suspected home loan fraud to authorities amid allegations that include AI-generated income statements. 

LATEST NEWS