How red teaming can help prevent healthcare fraud
Over the summer, federal authorities announced charges for 455 defendants as part of the 2026 National Health Care Fraud Takedown. The Department of Health and Human Services’ (HHS) Office of the Inspector General (OIG) alleges that the defendants perpetrated health fraud schemes involving more than $6.5 billion in false claims. But those cases were built after claims were submitted, and it’s unknown how much of that money can be clawed back.
Recent efforts have brought healthcare agencies closer to a proactive stance on addressing fraud, waste, and abuse. The Centers for Medicare & Medicaid Services (CMS) screens providers at enrollment through tools like the Advanced Provider Screening and initiatives like the Comprehensive Regulations to Uncover Suspicious Healthcare (CRUSH). But these defenses are built around known patterns, and program integrity leaders have few ways to test whether they hold up against schemes no one has seen yet. Especially with AI, schemes that can adapt quickly would be able to reach significant scale before a pattern emerges. By the time one is detected, the money is already gone.
HHS OIG, CMS, and other healthcare agencies need tools that can help anticipate and identify those threats more quickly. Red teaming—a common tactic used in IT security—is a practical strategy that agencies can use to probe for potential vulnerabilities that allow healthcare fraud, waste, and abuse to thrive.
Applying adversarial testing to healthcare fraud prevention
Red teaming is an authorized exercise that simulates how a realistic adversary might pursue a specific objective under carefully controlled conditions. The concept originates from IT security, where it tests whether an attacker could bypass safeguards and whether security teams would detect and contain the activity. The findings help organizations strengthen their defenses and test whether the improvements work.
ICF teams do this regularly in large-scale cloud development and deployment projects for federal clients. Our red teams design realistic scenarios, conduct controlled tests, and document both the weaknesses they validate and how system defenses respond. Findings lead to prioritized corrective actions and follow-up testing to verify the fixes worked.
The same principles can inform healthcare fraud, waste, and abuse prevention. Instead of probing a network, the simulation tests how bad actors might exploit healthcare benefit programs, payment systems, and enrollment processes, and assesses whether program controls recognize and stop the attempt.
Applying this approach to healthcare requires security testing expertise alongside a detailed understanding of program policy and operations. Enrollment requirements, coverage criteria, payment rules, and review procedures determine whether a simulated fraud scenario is credible. Combining security testing and healthcare domain expertise can help agencies test where safeguards might fail, identify practical improvements, and evaluate how those changes may affect legitimate providers and beneficiaries.
What would this look like in practice?
Using ICF Fathom, we can design and test a multi-agent architecture for healthcare agencies that red team different fraud, waste, and abuse scenarios. It could measure exposure and test potential controls, all without touching real claims or beneficiary data. Program integrity teams can then anticipate exploitation techniques virtually before they appear in reality. The tests would answer four questions:
What is vulnerable?
Agents submit synthetic applications and claims representing plausible fraud scenarios. The tests examine whether existing safeguards prevent an improper enrollment or payment.
How much is at risk?
For scenarios that bypass existing safeguards, the simulation documents how often the falsified information succeeds and where detection or intervention fall short. The analysis estimates potential financial exposure and identifies possible harm to patients.
What controls could close the gap?
Agents test additional or modified policy, data, system, and operational controls against the same scenarios, re-run at different levels of effort (e.g., low, moderate, high). Each test defines what the control should prevent.
How well could the controls work?
Comparing baseline exposure with residual exposure under each control assesses whether the changes interrupt the scheme. They also examine the effects on legitimate transactions, patient outcomes, and system burden.
Let’s apply that setup to a hypothetical example:
Imagine CMS wants to strengthen enrollment safeguards in durable medical equipment (DME) billing. In a red teaming exercise, the solution submits 100 synthetic applications alongside a set of legitimate applications to simulate the policy-based vetting of new DME suppliers. In 20 of those 100 falsified attempts, the information earns a DME billing credential.
Next, the exercise tests different policy, data, system, and operational controls. One of them, an additional validation of submitted application data through a third-party data source, cuts the fraudulent applicants’ success rate from 20% to 10%. Testing legitimate applications also reveals whether the safeguard introduces unnecessary denials, delays, or provider burden, and what additional costs would be incurred by using the third-party data source at scale.
Where sufficient evidence exists, analysts could use relevant historical data to estimate a range of potential exposure, distinguishing legitimate from potentially improper payments and making uncertainty explicit.
This red teaming exercise puts valuable information into CMS’ hands, but it doesn’t make decisions for them. It gives leaders evidence that they can use to choose a preventive response to fraudulent DME enrollment. Red teaming results can show how much a control would prevent, what it would cost legitimate providers, and where the remaining exposure exists, so CMS can choose the control that best balances protection against burden.
What credible healthcare red teaming requires
Applying red teaming to healthcare program integrity requires cybersecurity expertise, a secure testing environment, and a detailed understanding of healthcare policy and operations. ICF brings these disciplines together. Our red teaming and penetration-testing methodologies provide a structured foundation for defining objectives, designing realistic adversarial scenarios, documenting how safeguards perform, and retesting after improvements.
The scenarios must also reflect how healthcare programs actually operate. ICF’s policy, research, and clinical specialists can work alongside technical teams to design credible comparisons, establish defensible assumptions, and assess potential exposure. By testing synthetic cases and validating simulated decisions, agencies can identify changes that warrant further testing or a pilot.
Moving from reaction to action
Improving healthcare program integrity outcomes requires not only technology, but also a shift in mindset from reactive enforcement to preventive action. That’s what red teaming offers.
By simulating attacks in a controlled environment, HHS OIG, CMS, and other healthcare agencies can see where programs are exposed, estimate the potential impact, and test which controls could close the gaps before a real adversary finds them. Red teaming gives leaders something tangible to act on: data about which vulnerabilities are riskiest, which controls deliver the most protection, and how much burden each control adds to the system.
We already know how to test systems against cyber threats. We know how to analyze healthcare fraud patterns. The goal is simple: find the weakness before it can be exploited.