US Treasury AI tools prevent $4 billion fraud in 2024
The US Treasury Department has utilized machine learning systems to detect and prevent fraudulent payments worth billions of dollars in 2024. The Treasury issues checks for various federal programs, processing around 1.4 billion payments totaling $6.9 trillion annually. Their new method for detecting wrongdoers helped prevent and recover over $4 billion in fraudulent payments in the last fiscal year, six times more than in 2023.
Government agencies and financial institutions are increasingly turning to algorithms to identify fraudsters by using a variety of data on payment recipients to spot patterns associated with fraud. Despite the significant $4 billion figure recovered by the Treasury, it is still a fraction of the estimated total annual fraud losses by federal agencies, ranging from $233 to $521 billion.
Deputy Treasury Secretary Wally Adeyemo emphasized the importance of serving as stewards of taxpayers’ money and ensuring payments are made accurately. The Treasury’s efforts to combat fraud with algorithms aim to address biases present in historical data used for training fraud detection models, such as an overrepresentation of certain demographics in anti-fraud cases.
