Machine learning AI It helped detect $4 billion in fraud by 2024, the US Treasury said.

Machine Learning AI US Treasury Department Helped prevent and recover more than $4 billion in fraudulent or improper payments by 2024.

It represents a six-fold increase over the previous year, when the treasury received and earned $652.7 million.

Treasury Payment Integrity (OPI), part of the Office of Fiscal Services, says it has been using “machine learning AI” to identify high-risk transactions and quickly detect check fraud.

“Treasury takes seriously our responsibility to serve as effective stewards of taxpayers' money,” Treasury Deputy Secretary Wally Adeyemo said in a press release. “Helping agencies pay the right person, at the right amount, at the right time is at the core of our efforts.”

With online payment fraud expected to exceed $362 billion by 2028, Adeyemo said the Treasury will work with other federal agencies to provide tools and information needed to effectively combat fraud.

According to the U.S. Treasury, AI models have prevented $2.5 billion in fraud by identifying high-risk transactions, while Treasury check fraud detection systems have facilitated the recovery of $1 billion.

The press release said the Treasury implemented efficiencies in its payment processing programs, preventing $180 million in improper payments and expanding risk-based screening methods to prevent $500 million in fraud.

ELNA.ai founder Arun Prime Minister told Decrypt, “If AI systems are not properly designed and maintained, they are vulnerable to risks such as wear and tear, hallucinations and defects, which can lead to harmful outcomes.

“Prevention of fraud is a global issue, and the impact of AI can be greatly increased if the benefits are made more accessible to the public,” he said.

In May, the Treasury Department partnered with the Department of Labor to provide state unemployment agencies with access to the Unemployment Insurance Integrity Data Center's Do Not Pay Work system.

In the year Established under the Payment Integrity Data Act of 2019, the Do Not Pay System (DNP) is a centralized platform managed by the U.S. Treasury Fiscal Service designed to prevent improper payments by streamlining data verification for federal agencies.

Through the DNP portal, agencies can access multiple databases to verify recipient eligibility before sending funds. The agencies say it reduces the risk of errors, such as duplicate payments or payments to ineligible vendors.

The Treasury's AI push puts it alongside other federal heavyweights getting into the tech.

The IRS has been running machine learning models since last year to sniff out tax dodgers among the wealthy, NOAA has deployed neural networks to fine-tune hurricane forecasts by 2024 and provided real-time alerts in multiple languages.

Edited by Stacy Elliott.