The U.S. Treasury Department’s artificial intelligence system may begin to do what armies of accountants have long struggled to do: curb financial crime more effectively.
According to Treasury Department officials, a detection systemPowered in part by AI, it saved taxpayers $3.8 billion in its first year by freezing suspicious transactions and recovering ill-gotten payments. Now companies are trying to emulate this success, marking an important moment in the fight against financial fraud.
“With the right partners, the government will undoubtedly be successful in fraud prevention using AI – there are already many examples of such successes,” Shree TaylorVP Data Analysis and Innovation at Research on the elderlytold PYMNTS. “Other entities in the commercial and financial sector are also experiencing the benefits of implementing an AI-driven fraud detection system. The silver bullet is having the right workforce to develop and support a comprehensive fraud detection program.”
AI vs. Fraud
Under the new program, the Ministry of Finance now uses extensive screening to identify threats early and focus on high-risk transactions. It has added AI to detect check fraud faster and improved payment schedules to reduce losses.
Experts say private business owners can learn lessons from the Treasury Department’s example. AI-powered fraud detection systems can operate continuously and provide global coverage 24/7, 365 days a year, allowing businesses to quickly adapt to prevent financial losses and improve the security of commercial transactions. Vall HerardCEO of a risk management company Saifrtold PYMNTS.
“By incorporating both structured and unstructured data analytics, including sanctions lists, watch lists, Most Wanted lists, news articles, social media posts and customer communications, these systems can detect potential fraud indicators that conventional methods may miss,” he added.
Although AI can effectively prevent fraud, it is not always used. Herard said the financial industry has traditionally been conservative, with compliance teams often resistant to significant changes to their risk management systems.
“These systems, which have been developed and refined over the years, are rigorously tested and generally effective, leading to a cautious approach when it comes to adopting new technologies,” he added. “There is a prevailing uncertainty about how regulators might view the transition to AI solutions, which further dampens enthusiasm for rapid adoption.”
Herard said the Treasury Department’s success with AI in fraud prevention suggests that the government may have an edge over companies in some ways when it comes to using AI.
“This advanced position implies that regulators may be smarter than expected when evaluating AI technologies,” he added. “If a government organization demonstrates the effectiveness and reliability of AI in this critical area, it can significantly strengthen trust among financial institutions. This government seal of approval, coupled with regulators’ apparent sophistication in AI assessment, would likely make companies much more comfortable becoming fast followers, potentially accelerating the adoption of AI technologies in the trading and financial sectors .”
The future of fraud prevention
As technology advances, AI-powered fraud detection systems will likely improve their ability to analyze patterns and spot risks in real time while keeping users’ payments smooth, Herard said. These technologies can process more data and catch subtle fraud signals that people often miss.
“However, the future of AI in payment fraud detection is not just about more advanced algorithms,” he added. “It’s also about creating ecosystems where different AI models can work together seamlessly. For example, we could see the development of AI agents that can autonomously orchestrate complex fraud detection workflows, pulling in data from different sources and coordinating different analytical models as necessary. This approach could significantly increase the ability to detect and respond to emerging fraud tactics, improving efficiency and reducing the burden on human analysts.”
Shaun BarryGlobal Director of Risk, Fraud and Compliance at data and AI company SAS, told PYMNTS that fraud detection is becoming a differentiator for online payment systems.
“AI ensures that fraud detection occurs in near real-time and is part of every core operational payment system,” he added. “In fact, the companies that have the best fraud detection capabilities will be more profitable and have higher customer satisfaction than companies that fail to innovate. The best-in-class companies will go even further by using AI to assess customer friction related to fraud detection and balance the two to optimize the customer experience.”
Related: New AI system aims to detect financial fraud in corporate networks
