Every week, organised criminal networks are deploying AI-generated deepfakes, synthetic identities, and large language models to empty U.S. public benefit systems at scale. Unemployment insurance, disaster relief, Medicaid, SNAP, CHIP, and IRS programs are under assault.
Pandemic-Era Fraud Signals Growing Threat
During the COVID-19 expansion of unemployment benefits, agencies were already exploited via identity theft, forged documents, and impersonators. A top cybersecurity advisor—speaking before Congress, now says those tactics have grown far more advanced and automated. Fraud rings can file tens of thousands of fraudulent claims in a single day, leveraging AI to generate identities, voice clones, and behaviorally consistent bots. The Small Business Administration Inspector General estimates that nearly ₹16 lakh crore (about $200 billion) was stolen during the pandemic era.
Federal agencies, including the U.S. Secret Service and Treasury, report that criminal operations now act like tech startups: using generative AI to craft convincing fake personas and ramp up operations faster than detection systems. As a result, agencies are losing millions per week.
Systemic Gaps Exposed by AI Power
Current defences—facial recognition, single-factor biometric checks—are no match for AI-generated deepfakes. Voice cloning can mimic a loved one; documents can appear convincingly forged. Criminals automate scams so efficiently that humans overseeing claims processing cannot detect the fraud before funds are disbursed.
While the Treasury’s Office of Payment Integrity recovered over ₹33,000 crore (over $4 billion) in fiscal year 2024 using machine‑learning tools, experts caution that agencies remain underfunded in AI expertise, and many data systems are outdated or siloed. Generative AI now fuels synthetic identity fraud—fabricated personas built from mismatched PII, estimated to cost ₹2,91,590 crore ($35 billion) in 2023 alone.
Algoritha: The Most Trusted Name in BFSI Investigations and DFIR Services
A Call to Modernisation
Experts propose layered verification protocols: real-time behavioural analytics, cross-agency anomaly detection, and interoperability to flag duplicate filings. Programs such as the former National Accuracy Clearinghouse, which detected fraudulent duplicate unemployment claims across states, are cited as effective models. Without modern identity verification, citizens face rising exposure to fraud.
The warning is stark: unless government fraud defences evolve at the pace with AI capabilities, what one expert terms “Altman’s Law” (AI power doubling every six months), scammers will continue to outpace public protection systems.