Washington · AI utilization management
Washington is changing the standards governing AI-assisted utilization management and reinforcing human clinical accountability. Before you can understand what changes, you need to know how your payors behave today.
E2SSB 5395, Chapter 157, Laws of 2026, addresses transparency and accountability in prior authorization determinations, including the use of artificial intelligence when medical necessity is involved. Washington is establishing clearer requirements for human clinical judgment, transparency, and accountability when AI is used in prior authorization decisions involving medical necessity.
When AI is used for prior authorization functions involving medical necessity, the carrier must ensure the technology considers applicable individual clinical information and does not base its determination solely on a group dataset. AI criteria and guidelines must comply with applicable state and federal law, may not discriminate against an enrollee, and must be fairly and equitably applied. Policies and procedures governing AI use must be open to audit by the Washington Office of the Insurance Commissioner, carriers must periodically review AI performance, use, and outcomes, and patient data may not be used beyond its intended and stated purpose. These requirements also reach contracted entities performing applicable prior authorization functions.
The law also strengthens transparency around prior authorization determinations and restricts certain retrospective denials of previously authorized care. Prior authorization standards under RCW 48.43.830 take effect January 1, 2027.
Reporting expands beginning October 1, 2026
Washington itself is seeking greater visibility into how prior authorization decisions are being made. Providers should develop their own visibility into the payor behavior affecting their organizations. Unlock does not determine whether any payor is legally compliant.
New standards create a point of comparison. They do not automatically create visibility. If a health system doesn't understand existing behavior by payor, it may be difficult to determine whether utilization-management practices subsequently improve, remain unchanged, or shift elsewhere.
A health system can know that payor utilization management creates administrative burden without knowing precisely where that burden originates or how individual payors compare. Aggregate experience obscures significant differences: one payor may generate substantially more adverse medical-necessity decisions, another more requests for additional documentation, and appeal or overturn patterns may vary by service line.
The question isn't simply what the law requires. It's what your payors are doing today, and what happens next.
The goal is to understand variation in payor behavior, not simply to calculate an enterprise-wide denial rate. Depending on available data, analysis might include:
The value is identifying where the organization should pay attention. Differences among payors do not prove inappropriate AI use; they show where questions are worth asking with evidence behind them.
Use available organizational data to understand current payor behavior before attempting to evaluate what changes under the new standards.
Establish the baseline
Create a defensible picture of current payor utilization-management and denial behavior.
Compare your payors
Identify differences that enterprise averages may conceal.
Find the outliers
Determine where unusual patterns or disproportionate administrative burden warrant closer attention.
Monitor what changes
Compare future performance against the baseline as Washington's requirements take hold.
Put the evidence to work
Use the findings in managed care strategy, contract negotiations, payor discussions, escalation, and other appropriate channels.
Unlock has spent years examining payor utilization management, administrative burden, algorithm-assisted coverage decisions, and the appropriate role of AI in healthcare. Our work has consistently drawn an important distinction: AI can reduce administrative burden and improve healthcare operations, but human clinical judgment and accountability matter when technology influences whether patients receive care.
Article
Brandon Edwards examines how payors are using AI across prior authorization, denials, contract administration, and payment processes, and why providers need to understand how that technology may affect their financial performance.
Report · 2026
Our 2026 report examines the growing divide between AI that reduces administrative burden and AI used in coverage and denial decisions. It addresses human oversight, transparency, auditability, and accountability as regulators increasingly scrutinize how payors use AI.
Report · 2025
Our 2025 report documented emerging concerns around algorithm-assisted denials, including Cigna's PXDX system and UnitedHealthcare/NaviHealth, and argued that AI should not override clinical judgment or eliminate human review of individual circumstances.
Together, these resources establish a clear progression: payors are expanding their use of AI, concerns about AI-assisted coverage decisions have been building for years, and Washington is now establishing clearer expectations for accountability and human judgment.
For providers, the next question is practical: do you know enough about how your payors behave today to recognize what changes tomorrow?