US Treasury Uses AI to Recover $1.3bn in Tax Fraud in 2024

In the United States, the Treasury Department has increasingly utilized artificial intelligence (AI) to combat fraud, achieving significant results in fiscal year 2024. The agency reported halting $4 billion in improper payments, a substantial increase from the previous year. This included $1 billion in cheque fraud and $3 billion in other improper payments, facilitated by advanced machine-learning techniques that identify at-risk transactions. Treasury officials emphasized the transformative impact of AI in uncovering hidden patterns that fraudsters exploit. The Treasury's efforts are part of a broader initiative to ensure responsible management of taxpayer funds, as the agency oversees approximately 1.4 billion payments totaling over $6.9 trillion annually. The use of AI is complemented by human oversight, ensuring that final decisions on fraud labeling are made by personnel. This dual approach aims to enhance the accuracy and effectiveness of fraud detection. Additionally, the Internal Revenue Service (IRS) has also adopted AI technologies to address tax fraud, particularly targeting high-income individuals. Since late 2023, the IRS has recovered $1.3 billion from wealthy taxpayers, highlighting the agency's commitment to closing the tax gap, which is estimated to be around $496 billion annually. The IRS plans to increase audits using AI in the coming years. While the integration of AI in fraud prevention shows promise, regulators have raised concerns about potential risks associated with its use. Treasury Secretary Janet Yellen has called for responsible AI innovation, balancing efficiency gains with established risk management principles.