new york
CNN
—
The federal government’s bet on using artificial intelligence to fight financial crime appears to be paying off.
Machine learning AI helped the U.S. Treasury sift through reams of data and recover $1 billion worth of check fraud in fiscal year 2024 alone, according to a new estimate first shared with CNN. This is almost three times the amount collected by the Treasury in the previous financial year.
“It’s really been transformative,” Treasury Department official Renata Mikel said in a phone interview.
“Data has strengthened our efforts in detecting and preventing fraud,” Miskell said.
The Treasury Department credits AI with helping authorities prevent and recover a total of $4 billion worth of fraud in fiscal year 2024, a six-fold jump from the previous year.
U.S. authorities began covertly leveraging AI to detect financial crimes in late 2022, upending efforts already being made by many banks and credit card companies to thwart bad actors.
The goal is to protect taxpayers’ money from a surge in fraud during the coronavirus pandemic as the federal government rushes to provide emergency aid to consumers and businesses.
Indeed, Treasury is not using the kind of generative AI that has captivated users of OpenAI’s ChatGPT and Google’s Gemini by generating images, creating song lyrics, and answering complex questions. (Although I still sometimes struggle with simple queries.)
Instead, fraud detection efforts rely on machine learning, a subset of AI that excels at analyzing vast amounts of data and making decisions and predictions based on what it learns.
AI can be extremely useful in fighting financial crime by sifting through nearly endless data streams and detecting subtle patterns. All of this can be done in a fraction of the time it would take a human to do it. Experts say that once advanced AI models are trained, they can sniff out suspicious transactions in just milliseconds.
“Scammers are very good at hiding. They’re secretly trying to game the system,” Miskell said. “The use of AI and data can help discover and prevent these hidden patterns and anomalies.”
This is especially important for the Treasury, which is one of the largest payers on the planet, if not the largest.
Each year, the Treasury Department provides approximately 1.4 billion payments worth approximately $7 trillion to 100 million people. It is responsible for sending everything from Social Security and Medicaid payments to federal employee paychecks, tax refunds and stimulus checks.
Because of its critical role, the Treasury Department is a prime target for fraudsters looking to steal from taxpayers.
Last year, the Internal Revenue Service announced it was deploying AI to detect tax fraud by examining large and complex returns from hedge funds, law firms, and others.
Juniper Research estimates that online payment fraud is expected to exceed $362 billion by 2028.
And some of that fraud is fueled by AI itself.
In one infamous case earlier this year, Hong Kong police said a financial official was tricked by a deepfake video into paying a scammer $25 million.
U.S. officials have expressed concern that AI poses new risks to the financial system. Treasury Secretary Janet Yellen warned bankers in June that AI in the financial industry poses “significant risks.”
Yellen’s top regulators late last year classified AI as an “emerging vulnerability” to the financial system.
Mikel stressed that while the AI system will alert on suspicious transactions, “humans are always watching,” and the final determination of whether something constitutes fraud will be made by federal agencies.
The Treasury Department’s use of AI to fight financial crime is only in its infancy.
Mikel indicated that the Treasury Department is considering how to implement fraud detection methods used by major banks and credit card companies, but is not giving details to avoid “bad actors snitching.” I didn’t go into it.
A Treasury Department spokesperson told CNN that the department is accelerating efforts to strengthen fraud detection tools available in federal and state-administered programs. Officials are testing new data sources to better spot fraud and suspicious payments, and are working with state agencies to combat unemployment insurance fraud.
