Fighting Fraud at Machine Speed

Presented by Socure & Carahsoft

Anderson opened by identifying one of the most persistent structural challenges facing federal agencies: the absence of a shared signal framework. When agencies operate with siloed technology and siloed processes, Agency A may understand a fraud pattern that Agency B has never seen — and by the time the knowledge spreads, millions of taxpayer dollars may already be lost. The solution, Anderson argued, is not necessarily a shared process, but a shared ecosystem of understanding — one where intelligence, not just data, flows across organizational boundaries.

Central to that vision is what Anderson called a "data is power" mindset. For too long, agencies treated data as a liability — something to be minimized and guarded. But in the fight against fraud, more data means better anomalous analysis, better signal detection, and faster response. Paired with modern application development and effective data science programs, agencies can move from months-long change control cycles to near-real-time fraud remediation — critical when fraudsters are extracting millions of dollars a day.

The most urgent challenge Anderson sees ahead, however, is the evolution of the fraud threat itself. Today's fraudsters are already using AI to spoof locations, forge documents, and defeat human reviewers. Tomorrow, those tools will be even more sophisticated — capable of generating fake documents so convincing that no human in the loop will catch them. Anderson's prescription: match AI with AI. Move from human-in-the-loop identity validation to automated machine learning models specifically designed to detect and defeat generative AI-powered fraud.

Key Takeaways:

• A shared signal framework — enabling intelligence, not just data, to flow across agency boundaries — is essential to fighting fraud at scale rather than one program at a time.
• The "data is power" mindset shift is foundational: more available data enables better anomaly detection and faster fraud response, reducing taxpayer losses.
• The next generation of fraud will be AI-driven — agencies must move from human-in-the-loop validation to automated machine learning models capable of defeating generative AI fraud techniques.