Payers Are Using AI Against You. Here's What It's Costing You Right Now.

If you work in healthcare finance and you have been waiting for a clear sign that the AI arms race is real and that it is already costing you money, this was your week.

Three things happened in the span of about 72 hours that, taken together, paint a pretty uncomfortable picture for health system CFOs and revenue cycle leaders. None of them got the attention they deserved. So let's go through them one at a time and then talk about what you're actually supposed to do with this information.

First: UnitedHealth is spending $3 billion on AI. You should care a lot about that.

STAT News published a detailed examination this week of UnitedHealth Group's AI push, and the numbers are not small. UHG currently employs 22,000 software engineers worldwide. More than 80 percent of them are now using AI to write code or build new agents, a significant jump from just a few years ago.

Here is the part that matters for RCM leaders: UHG's stated goal is to use AI to speed up decision-making and streamline the claims and billing machinery. That includes automating fraud detection, clinical documentation, and the selection of billing codes that determine how much a given encounter costs and who pays for it.

Read that last sentence again.

The largest payer in the country is using AI to automate the process of deciding which billing codes are appropriate and how much to pay. They have 22,000 engineers working on this. Your revenue cycle team is how large exactly?

This is not a hypothetical future risk. This is a description of what is already happening at scale on the payer side of every claim you submit.

Second: Payer take-backs are accelerating. Most providers have no idea how much they're losing.

Waystar CEO Matt Hawkins gave an interview this week that included a data point that should be on every CFO's radar.

Payer recoupments, meaning payments already received and recorded as revenue that then get clawed back by insurers, are costing healthcare providers more than $1.6 billion every month. And recoupments are growing at twice the rate of overall claim volume over the last three years.

Twice the rate.

Hawkins called these "silent denials" because, unlike a standard denial that triggers a workflow, recoupments often arrive months or even years after a claim was initially adjudicated and paid. They show up buried in remittance data. Most providers lack the visibility to even identify which claims were affected, let alone why or how to appeal.

Waystar built an AI tool specifically to address this, and they shared early results from an unnamed health system with roughly $4 billion in annual revenue. That system identified $32 million in hidden recoupments using the tool. Work that would have required the equivalent of 27,000 hours of manual reconciliation and 13 full-time employees.

Think about that number for a moment. Thirty-two million dollars sitting in remittance data that nobody caught. Not because the team was incompetent. Because this problem was designed to be invisible.

I wrote in RCM 2030 Strategy and Survival that payers would increasingly use data and automation as leverage in the provider-payer relationship, and that CFOs who failed to build equivalent analytical capability would find themselves at a structural disadvantage. The recoupment problem is exactly that dynamic playing out in real time. Payers have the data infrastructure to identify and execute these adjustments quietly. Most providers do not yet have the infrastructure to catch them.

Third: The federal government just added a new enforcement layer on top of all of this.

Acting U.S. Attorney General Todd Blanche signed a memo this week formally launching the National Fraud Enforcement Division, a new DOJ unit specifically focused on taxpayer-funded programs. The division includes a dedicated Healthcare Fraud Unit, and every U.S. Attorney's Office has been directed to appoint a prosecutor accountable to it within 21 days.

The memo stated that more than $1 trillion is at stake annually. The same week, the Texas Attorney General launched investigations into dozens of Medicaid providers. Federal scrutiny of Medicaid specifically has been accelerating across at least 10 states.

Here is the revenue cycle implication that most coverage of this story missed: this enforcement environment does not exist in a separate lane from the AI arms race. It exists in the same lane.

Payers under pressure to manage their medical loss ratios are using denials and recoupments more aggressively. They are also feeding data to a federal oversight apparatus that is now significantly more resourced and more focused than it was 30 days ago. If your coding, documentation, and billing processes have gaps, the probability that those gaps get surfaced has just increased meaningfully. Not because your team is doing anything wrong. Because the scrutiny infrastructure around you just got bigger.

So what does this actually mean for your revenue cycle right now?

It means three things worth acting on.

The first is that recoupment visibility is no longer optional. If you do not have a process for systematically identifying and tracking payer take-backs, you are almost certainly losing money you do not know about. The Waystar data suggests this is not a minor rounding error problem. It is a structural revenue leak. Whether you use Waystar's tool or build the capability internally or through another vendor, you need line of sight into recoupment activity before it becomes a write-off pattern.

The second is that your coding and documentation practices need to be audit-ready right now, not when something goes wrong. The combination of AI-assisted payer adjudication and a newly empowered federal fraud enforcement apparatus means that the bar for documentation integrity has effectively risen. This is not about paranoia. It is about the entirely predictable consequence of a more automated, more scrutinized claims environment. Clean claims, accurate coding, and solid documentation are your best protection.

The third is that AI on your side of the equation is no longer a nice-to-have. I understand the budget pressure. I understand the competing priorities. But UHG has 22,000 engineers using AI to manage the process of adjudicating your claims. The information asymmetry between payers and providers on this front is growing, not shrinking. The organizations that close that gap by building analytical and AI capability on the provider side will have better margins, fewer surprises, and more successful appeals. The ones that wait will keep absorbing losses they do not fully understand.

This is not a 2030 problem. It is a right now problem. The arms race has already started, and one side is significantly better armed than the other.

The good news, if you want to call it that, is that the tools to even the playing field are available. The Waystar story this week is one example. The question is whether your organization is moving fast enough to use them.