CMS is using AI to audit
hospice and home health billing.

Two independent systems now act on the same hospice billing decisions. Software vendors sell tools that recommend eligibility, recertification, and coding decisions. CMS, separately, uses analytic software to flag billing that resembles fraud. Because both operate on the same decisions, an agency that follows its vendor's recommendations can be steered toward the patterns CMS is built to detect.

Your software vendors are putting AI into real decisions, including eligibility, recertification, length of stay, OASIS coding, and visit narratives. Some agencies have bought predictive tools that recommend who qualifies for care and for how long. Most of these tools are not interpretable: the agency cannot fully explain how a recommendation was reached, and no policy governs the tool's use.

As of 2026, the use of AI in the eligibility decision is no longer hypothetical. Hospice EMR vendors have begun shipping AI features that review recertification eligibility and terminal-diagnosis documentation inside the clinical workflow, before a nurse signs off. No independent party has checked whether these tools push toward the same billing patterns CMS is built to flag.

At the same time, CMS is using AI and predictive analytics to find billing that resembles fraud. The patterns it scores are the same ones a predictive tool can move an agency toward. An agency that aligns its billing to a vendor's model may therefore move toward the exact patterns CMS is built to detect. A vendor recommendation does not, by itself, satisfy the documentation an auditor requires to justify a recertification.

These conditions create a gap: vendors sell the tools, no professional association has issued standards for their use, and CMS reviews the decisions those tools influence.

This enforcement is already live

This enforcement is already underway. The following developments occurred over the past year.

• • A nationwide enrollment moratorium on new hospice and home health agencies took effect in May 2026. It runs six months and can be extended. Existing providers continue to bill, so for an established agency the moratorium chiefly indicates heightened enforcement scrutiny. CMS is actively screening for these billing patterns.
• • CMS stood up a Fraud Defense Operations Center that uses AI and machine-learning models to flag claims for immediate audit. CMS reports $1.8 billion in payments suspended in 2025.
• • CMS reports $11.9 billion in potentially fraudulent payments prevented across fiscal years 2022 through 2024. Automated prepayment denials accounted for $132 million of the fraud-related total across those same years.
• • The largest health care fraud takedown on record came in June 2025. It named 324 defendants and alleged $14.6 billion. CMS prevented more than $4 billion from being paid and suspended or revoked the billing privileges of 205 providers.
• • In four high-oversight states, 122 hospices had their billing privileges revoked as of June 2025.
• • OIG has found that roughly one-third of general inpatient hospice claims were billed in error.
• • In June 2026, an OIG review found that 45 of 100 sampled new-enrollee certification periods did not meet Medicare requirements, and that 21 of them lacked clinical evidence of a terminal illness. OIG estimated Medicare could have saved a projected $255.1 million.
• • In late 2025, CMS ran a public competition to build explainable AI that scores hospice claims for fraud.

A note on these numbers. The largest dollar figures are CMS's projected cost-avoidance estimates, not collected dollars. The moratorium blocks new enrollments and ownership-change applications, so a long-standing agency is not frozen out. These figures describe a functioning enforcement apparatus directed at hospice and home health billing.

What CMS scores you on

CMS publishes a hospice scorecard called the Service and Spending Variation Index. It is built on claims patterns: long length of stay, live-discharge churn where a patient is discharged and then readmitted to the same hospice within a week, curative spending while a patient is on hospice, and visit patterns across the week.

For home health, CMS applies analogous criteria: OASIS coding that shifts patients into higher-paying groups, therapy utilization that diverges from what the patient's condition predicts, and visit counts calibrated to reach a payment threshold.

A predictive tool designed to optimize any of these metrics will, by design, move an agency's billing toward the patterns CMS monitors. The agency, not the vendor, bears the resulting liability.

Why a black-box tool is hard to defend

Two tools trained on the same patient data can disagree. A model trained on clinician certifications tends to predict eligibility, while a model trained on auditor denials tends to restrict it. The same inputs produce opposite recommendations, and the agency owns whichever answer it billed on.

The harder problem is defensibility. Say AI informed a recertification, and the agency cannot explain the clinical basis. There is no model documentation, no record of where the data came from, and no human reconciliation, so there is nothing to put in front of a reviewer. If a tool cannot explain a recommendation, the agency cannot defend it.

The defensible position

Your existing obligations already cover this. Under OIG's General Compliance Program Guidance, every agency owes the seven elements of an effective compliance program: a written policy, oversight, training, auditing and monitoring, and prompt correction. Those same requirements apply to the AI tools you and your vendors are already running. This is not a new regulatory regime, and it does not require buying more software.

In practice, this approach comprises four measures: (1) a written policy treating AI output as one input rather than the determinative factor in eligibility or coding; (2) a requirement that a clinician make and document the clinical basis for each decision; (3) an inventory of the tools that affect any billable decision; and (4) periodic comparison of the agency's billing against the patterns CMS scores, so deviations are identified before an analytic model flags them.

Where I fit

I am a PhD clinician who has worked inside interdisciplinary care teams. I help hospice, home health, and palliative care agencies put governance around the AI they already use, written for the people who have to follow the policy. Agencies seeking to establish governance for their existing AI tools can contact me to discuss their situation.