Payer AI

Payer AI covers the AI systems on the payer side of the U.S. healthcare economy — the tools health plans, PBMs, MACs, and (increasingly) hospital revenue-cycle teams use to review claims, run prior authorization, adjudicate coverage, and manage utilization. It has been the highest-friction corner of healthcare AI: legislators, regulators, and clinicians have all taken direct aim at it, and 2026 has been a busy year on both sides.

Why this got contentious fast

Two things happened at once. The first is that health plans quietly deployed ML-driven prior-authorization systems at scale — some of them making initial denial decisions on hundreds of thousands of requests without human review at the first pass. Investigative reporting through 2023–2025 (STAT, ProPublica, HHS-OIG audits) surfaced enough class-action-worthy pattern-of-denial cases that Congressional oversight followed. The second is that generative AI made it dramatically easier for hospital revenue-cycle teams to submit more prior-authorization requests, on average better-supported, at higher levels of service. The PwC health-plan cost-inflator analysis projected this as a material driver of medical-loss-ratio pressure into 2027.

The result is an AI arms race with a policy layer sitting on top of it.

The 2026 rule stack

• CMS’s Interoperability and Prior Authorization Final Rule. Effective through 2026–2027, this rule forces Medicare Advantage, Medicaid managed-care, and other CMS-regulated plans to shorten decision windows, publish denial rates, and expose electronic prior-authorization APIs. Every plan has an ML system somewhere in that pipeline; the rule reshapes both the incentives and the audit surface.
• State-level PA reform. A growing patchwork of state laws (Texas, California, several others) restricting fully-automated denials and requiring physician review before adverse determinations.
• Federal action on algorithmic bias in coverage decisions. ACA §1557 non-discrimination requirements now explicitly reach automated decision tools used in coverage determinations.

Where the money is going

On the vendor side, capital is flowing into three buckets:

1. “Prior-auth automation for providers” — companies that use AI to prepare and submit PA requests on behalf of hospitals and clinics, minimizing physician time.
2. “Prior-auth intelligence for payers” — the counter-play: better ML for adjudication, coverage-policy management, and denial-rationale drafting.
3. “Revenue-cycle automation” — the broader wrapper around all of it: coding, charge capture, denials management, appeals drafting.

The interesting question — and the one we return to often in our reporting — is what the equilibrium looks like when both sides are augmented. Does it net out to more accurate care, or just an escalation of overhead?

Where the litigation is running

Payer-AI litigation has become a category of its own. Several class-action suits filed 2023–2025 targeted Medicare Advantage and Medicaid managed-care plans over algorithmic denial patterns, with plaintiffs pointing at consistent under-denial-rationale evidence and disparate impact on protected classes. Settlements have varied but a common pattern is emerging: undertakings to publish denial-rate metrics, restrict fully-automated adverse determinations, and submit the underlying models to third-party audit. Watch this litigation trend closely — it is doing as much to shape payer AI practice as the formal rulemaking.

What we cover

Our reporting in this topic focuses on: CMS rulemaking on prior authorization, state-level regulation of automated coverage decisions, litigation and settlements involving algorithmic denials, vendor category maps on both sides of the transaction, and the practical mechanics of how prior-authorization AI is deployed inside health plans and hospital revenue-cycle teams. Explore related articles and news below.