News built for finance pros
CFO Brew helps finance pros navigate their roles with insights into risk management, compliance, and strategy through our newsletter, virtual events, and digital guides.
In a world without fraud, generative AI could just be used to design cursed Valentine’s Day candies or write school absence notes in the style of Moira from Schitt’s Creek.
Sadly, we do not live in that world, and anti-fraud teams see plenty of opportunity to make use of the rapidly developing tech. More than four in five organizations say they’ll start using generative AI to detect fraud in the next two years, according to a new survey from the Association of Certified Fraud Examiners. (The ACFE partnered on the report with analytics company SAS, which sells an AI and analytics platform.)
Of course, 83% of organizations planning to use gen AI within two years doesn’t guarantee 83% will actually make it happen by then. In surveys from 2019 and 2022, respondents also said they expected to start using more of “every data analysis technique in our study,” but “reported adoption rates have shown little growth since 2019, highlighting the slow pace at which organizations are able to implement new technologies.”
To be a little more optimistic, AI isn’t totally new to some anti-fraud teams. Almost one in five (18%) are already using AI and machine learning tools, the broader category to which gen AI belongs. That’s a five-percentage-point increase from 2013.
Anyway, back to pessimism. Generative AI sometimes makes stuff up, and the stakes in anti-fraud work are a bit higher than when we ask ChatGPT about a historical event we were too lazy to look up on Wikipedia. “Investigators will find it difficult to trust or deploy an inaccurate technology,” one respondent said, calling accuracy “the biggest challenge for generative AI.”
Meanwhile. A type of technology that more fraud detection teams are using, even if the tech is generative of fewer headlines, is biometrics. Four in 10 said their anti-fraud measures include physical biometrics like fingerprints, voices, and facial features, up from 26% in 2019.