For years, conversations about government fraud have focused on scale. The numbers are staggering. The Government Accountability Office estimates between $233 billion and $521 billion is lost to fraud annually but if we say the quiet part out loud that estimate is far too low. The reality that emerged from a recent cross-government roundtable of federal leaders shines a light on the urgency in this moment that is greater than any single number.
Fraud is no longer just large. It is fast.
Today’s fraud landscape is defined by speed, adaptability, and coordination at a level government systems were not built to match. What once took months to orchestrate can now happen in days, or even hours. As detailed in the report from the roundtable, in one documented case, fraud actors used artificial intelligence to generate more than 24,000 synthetic identities and launch nearly 36,000 attacks within a matter of weeks, many within just 48 hours of identity creation. That velocity makes it increasingly hard to keep up with the fraudsters.
This is no longer a question of fixing gaps in existing systems. It is a structural mismatch between how quickly fraud evolves, how slowly public sector controls adapt, and how the public sector continues to implement yesterday’s approaches to stopping fraud to a next generation problem. Programs designed to update annually by looking backwards are facing threats that iterate in real time and not as slowly as the federal budget process.
At the same time, something else is happening beneath the surface. The signals the government has long relied on to detect fraud are weakening. Identity markers, like consistent email addresses, phone numbers, and device fingerprints, are disappearing as fraudsters deliberately engineer around them. The result is a system where traditional safeguards are not just failing, they are being outpaced. And yet, the most striking takeaway from the roundtable was not the risk. It was the opportunity to do something better.
Prevention Is Finally Within Reach
There is now clear, operational evidence that fraud can be prevented before it happens. Recent efforts across Treasury and oversight bodies show that earlier intervention, backed by better data and modern analytics, can stop improper payments upstream without slowing down services. Prevention and speed are no longer opposing forces. Done right, they reinforce each other.
For decades, the government has relied on a “pay and chase” model. Funds go out quickly, and fraud is identified, prosecuted, and recovered later. But recovery is expensive, slow, and often incomplete. Prevention, by contrast, is immediate and far more effective. The challenge is that government incentives have not fully caught up.
Today, agencies are still rewarded for rapid disbursement and post-payment recovery, even as practitioners and evidence alike show that earlier screening delivers better outcomes. Fixing that misalignment is fundamentally a policy challenge.
The Coordination Problem
That theme emerged most sharply in discussions around federal grants. It is clear there are tools the public sector has available to prevent fraud, some older and many that have yet to be adopted. However, the responsibility, data, and intelligence is spread across agencies, states, and systems that were never designed to work together. When billions of dollars move through decentralized structures, even small coordination gaps can create significant exposure. Closing those gaps requires orchestrating what already exists, not reinventing the system.
And there are real reasons for optimism. Data-sharing authorities are expanding. Tools like Treasury’s Do Not Pay system are proving their value, identifying and preventing millions in improper payments with strong returns on investment. (This strategy is only scratching the surface of what is needed.) Perhaps most importantly, there is growing bipartisan consensus that prevention must happen earlier in the payment lifecycle. The Administration’s recent Executive Order on Fraud demonstrates commitment at the highest levels.
The Window to Act
The question now is whether that momentum can be sustained and scaled. Because fraud actors are not waiting. They are already adapting to new controls, leveraging artificial intelligence, global infrastructure, and increasingly sophisticated identity strategies. The window to respond is now measured in days or months, rather than years.
The takeaway from this roundtable is that the path forward is clearer than ever.
- Move prevention upstream before the payment goes out.
- Treat data and fraud risk intelligence sharing as essential and find the right way to do it.
- Align incentives with outcomes that reward dollars never lost, not just dollars recovered.
Fraud is evolving quickly. The real test is whether government can evolve just as fast.
Jordan Burris is the Head of Public Sector at Socure, an industry-leading digital identity verification provider. Burris was previously Chief of Staff to the Federal CIO at the White House during the first Trump Administration.
