AI Regulation Tracker / State law in force
Arizona Now Requires a Human Medical Director to Review Medical-Necessity Denials, Barring AI-Only Rejections
As of July 1, 2026, Arizona's House Bill 2175 is in force. Before a health insurer can deny a claim or a prior authorization on medical-necessity grounds, its medical director has to review the denial personally and cannot rely solely on recommendations from any other source. In practice, that means an algorithm cannot be the last word on turning you down.
Arizona House Bill 2175 has been on the books since Governor Katie Hobbs signed it on May 12, 2025. The reason it matters this week is the effective date. The act carries a clean trigger, "This act is effective from and after June 30, 2026," which puts it into force on July 1, 2026. The waiting period is over. The human-review requirement is now something insurers doing business in Arizona have to actually do.
What is now in force
The core of the law is a checkpoint. Before a health-care insurer can deny a claim or a prior authorization, and the trigger is a denial on grounds of medical necessity, experimental or investigational status, or any matter that requires medical judgment, the insurer's medical director has to step in. The statute says the medical director must individually review the denial, must exercise independent medical judgment, and may not rely solely on recommendations from any other source.
Read that last phrase carefully, because it is the point. "May not rely solely on recommendations from any other source" is drafted broadly. It sweeps in an outside reviewer, a vendor, a checklist, and, most relevant right now, an automated or AI-driven utilization-review tool. The medical director can look at what the algorithm flagged. What the medical director cannot do is let the algorithm be the whole basis for the no. A licensed human has to review the file, apply independent judgment, and put a name on the denial.
HB 2175 builds this requirement into two places in the code. It creates ARS §20-3103 on the claims side and ARS §20-3407 on the prior-authorization side, so the same human-review discipline applies whether the insurer is refusing to pay a submitted claim or refusing to preapprove care in advance. The Arizona Department of Insurance and Financial Institutions is the agency that enforces it.
What the law does not say
I want to be careful here, because the easy headline gets this wrong. Arizona did not ban AI in claims handling. Nothing in HB 2175 tells an insurer it cannot use an algorithm, a model, or an automated tool in its utilization-review process. Insurers can keep using those tools to sort, flag, and triage.
What the law does is narrower and more precise. It says the tool cannot be the sole basis for a medical-necessity denial. There has to be a human medical director in the loop, individually reviewing the denial and exercising independent medical judgment, and that person cannot lean entirely on the machine's recommendation. If you describe this as an "AI ban," you are overstating it and you will misjudge what compliance actually requires. The requirement is a mandated human checkpoint on denials, not a prohibition on software.
Why this is happening now
Arizona is not acting alone. This law is part of a visible multistate wave aimed at the same worry, which is automated systems generating high volumes of claim and prior-authorization denials with little or no meaningful human review. Iowa moved with House File 2635. Indiana moved with House Bill 1271. Washington moved with Senate Bill 5395. The through-line across all of them is the same instinct Arizona wrote into HB 2175: keep a qualified human accountable for a denial that affects someone's care.
That pattern is worth watching if you operate in more than one state. The specific statutory language differs, but the direction of travel is consistent. States are drawing a line at the denial, and the line requires a person, not a model, to own the medical judgment.
What this means for insurers, vendors, and providers
For health insurers and utilization-review entities operating in Arizona, the compliance question is concrete. Can you demonstrate that a medical director individually reviewed each medical-necessity denial and exercised independent judgment, rather than rubber-stamping an automated output. If your denial workflow routes cases through an AI tool and out the door with a signature that never involved real review, that is the exposure this law is built to catch. The fix is process and documentation: a genuine human review step, and a record that shows it happened.
For the insurtech and AI utilization-review vendors selling into this market, the product implication is a design one. A tool that positions itself as automating denials at scale is now selling something an Arizona insurer cannot lawfully use as the sole basis for a medical-necessity no. The tools that fit the new rule are the ones that support a human reviewer, surface the record, and make the medical director's independent judgment easier to exercise and easier to prove, rather than the ones that replace that judgment.
For physicians and provider billing teams, this is a practical lever. If an Arizona health insurer denies care on medical-necessity grounds, the law now requires that a medical director individually reviewed it and did not rely solely on some other source. That gives providers a clearer basis to press on a denial that looks automated and to ask whether the required human review actually occurred.
What to do now
If you run claims or utilization review in Arizona, confirm your denial process has a real medical-director review step in front of every medical-necessity, experimental or investigational, and medical-judgment denial, and confirm that step is documented well enough to survive a look from the Department of Insurance and Financial Institutions. If you build or buy AI utilization-review tools, check that they are configured to inform a human decision rather than to be the sole basis for one. If you are a provider on the receiving end, know that the human-review requirement is now in force and can be raised when a denial looks machine-made. And if you operate in multiple states, treat Arizona as one node in a wider pattern and read your own states' versions rather than assuming they match.
Questions professionals are asking
Does Arizona HB 2175 ban AI in health insurance claims?
No. The law does not prohibit insurers from using AI or algorithmic tools in utilization review. It requires that a medical director individually review a medical-necessity denial and may not rely solely on recommendations from any other source, which means AI cannot be the sole basis for the denial. The tool can inform the decision, but a human has to own it.
When did the law take effect?
July 1, 2026. Governor Katie Hobbs signed HB 2175 on May 12, 2025, but the act carries the clause "This act is effective from and after June 30, 2026," so the requirement became operative on July 1, 2026.
What exactly does the medical director have to do?
Before a health-care insurer denies a claim or a prior authorization on medical-necessity, experimental or investigational, or medical-judgment grounds, the insurer's medical director must individually review the denial, exercise independent medical judgment, and may not rely solely on recommendations from any other source. The requirement is written into ARS ยง20-3103 for claims and ARS ยง20-3407 for prior authorization.
Who enforces it, and who does it apply to?
The Arizona Department of Insurance and Financial Institutions enforces it. It applies to health insurers and utilization-review entities operating in Arizona, and it has direct practical relevance for physicians, provider billing teams, and the insurtech and AI utilization-review vendors that sell into that market.
Is Arizona the only state doing this?
No. It is part of a broader wave. Iowa (House File 2635), Indiana (House Bill 1271), and Washington (Senate Bill 5395) have moved on the same concern about automated claim and prior-authorization denials. The statutory language varies by state, so read each one on its own terms rather than assuming they match.
RELATED BRIEFINGS
Browse the full AI Regulation News tracker
Informational analysis for working professionals, not legal, insurance, or medical advice. Confirm how any statute or requirement applies to your situation with qualified professionals in the relevant jurisdiction.