AI & Data Exchange 2026: PRAC’s Ken Dieffenbach on Using AI Tools to Stay a Step Ahead of Fraudsters

Why It Matters

AI‑driven oversight gives the government a scalable way to protect taxpayer dollars while maintaining rapid fund distribution, setting a new standard for fraud prevention in large‑scale relief programs.

Key Takeaways

• PRAC’s AI engine reviews 20,000 applications per second
• Over 1,000 cases involve $2.5 billion in potential fraud losses
• Simple yes/no data checks identified 1.4 M invalid SSNs
• Fraudsters now use AI to forge realistic documents

Pulse Analysis

The Pandemic Response Accountability Committee, originally created under the CARES Act to monitor emergency COVID‑19 spending, has been reauthorized through 2034 by the One Big, Beautiful Bill Act. This legislative boost keeps PRAC in charge of a staggering $5 trillion portfolio, but the sheer volume of funds demands more than traditional audit methods. By integrating artificial‑intelligence, PRAC can ingest and analyze massive datasets in real time, turning what was once a labor‑intensive "pay and chase" approach into a proactive defense against fraud.

At the heart of PRAC’s new strategy is an AI‑enabled fraud‑prevention engine trained on over five million pandemic‑relief applications. Capable of processing 20,000 submissions per second, the system flags anomalies before money leaves the Treasury, allowing grant officers and contracting officials to intervene early. Coupled with streamlined yes/no validations against Treasury’s Do No Pay files and the Social Security Administration’s Death Master File, PRAC has already uncovered 1.4 million questionable Social Security numbers linked to $79 billion in disbursed funds. These data‑driven safeguards demonstrate that speed and integrity can coexist in federal spending.

However, the arms race intensifies as fraudsters adopt the same AI technologies to generate convincing fake documents and overwhelm weak program controls. Dieffenbach warns that auditors must continuously evolve their tools to detect synthetic receipts, contracts, and other forged artifacts. The ongoing battle underscores the need for collaborative data sharing, robust machine‑learning models, and a culture of pre‑emptive risk assessment across agencies. As the government refines its AI capabilities, PRAC’s experience offers a blueprint for other oversight bodies seeking to protect billions of taxpayer dollars in future emergency relief efforts.