Your insurance company might be using artificial intelligence right now to decide your premiums, approve your claims, or even deny your coverage. But here’s the problem: there’s no consistent rulebook telling them how to do it fairly.
The National Association of Insurance Commissioners (NAIC) tried to fix this in December 2023 with an AI model bulletin outlining expectations for insurers using AI. The catch? Only 24 out of 56 jurisdictions have adopted it as of August 2025, and regulators remain deeply divided on whether more rules are needed.
If you’re wondering why your health insurance claim got auto-rejected or your car insurance premium suddenly jumped after an algorithm review, this regulatory gap might be why. Let me break down what’s actually happening behind the scenes.
What the NAIC AI Bulletin Actually Requires (And What It Doesn’t)
The NAIC’s December 2023 bulletin sounds impressive: insurers must ensure AI systems comply with existing laws against discrimination, maintain human oversight, and document how algorithms make decisions.
But here’s what it doesn’t mandate:
- No standardized disclosure format exists across states. What Colorado requires for AI transparency differs wildly from what New York demands—or what 32 states don’t require at all.
- Zero penalties specified for violations in the model bulletin itself. Enforcement depends entirely on each state’s existing insurance laws, which weren’t written with AI in mind.
- The “evaluation tool” the NAIC proposed? Still under debate. The 60-day comment period ended September 5, 2025, with no consensus on whether it should assess AI risks or if that’s regulatory overreach.
Translation: Your insurer’s AI might follow the bulletin’s principles, but there’s no uniform way to verify it or punish violations. That’s not exactly consumer-friendly.
4 States That Aren’t Waiting for National Standards
While the NAIC debates, four states jumped ahead with their own AI insurance rules:
| State | Key AI Requirement | Enforcement Status |
|---|---|---|
| Colorado | Annual AI impact assessments required | Active since 2024 |
| New York | Algorithmic bias audits for underwriting | Phased rollout through 2026 |
| California | Consumer right to know if AI denied claim | Proposed (under review) |
| Texas | Human review for all AI-flagged claims over $10,000 | Effective Q1 2025 |
These states recognized a critical problem: AI systems trained on historical data can replicate past discrimination. If insurers historically charged certain ZIP codes higher rates due to redlining, an AI trained on that data will do the same—unless someone forces transparency.
The Texas rule about human review for claims over $10,000 matters because it creates a clear consumer protection mechanism. If your medical claim for $15,000 gets auto-denied, a human must actually look at it. But if you live in one of the 32 states without AI-specific guidance? No such guarantee exists.
Why Regulators Can’t Agree on More AI Rules
The NAIC split breaks down into three camps:
Camp 1: “Existing laws already cover this.” Some regulators argue that anti-discrimination statutes, unfair trade practice laws, and underwriting standards already apply to AI decisions. Adding new rules just creates compliance costs without improving consumer outcomes.
Camp 2: “We need AI-specific transparency now.” Other regulators point to the “black box” problem—even insurers don’t always understand why their AI makes certain decisions. How can you prove discrimination if you can’t audit the algorithm?
Camp 3: “Let’s wait and see what goes wrong.” A smaller group wants data on actual AI-related consumer harm before imposing requirements. Problem is, consumers can’t report harm they don’t know exists.
Here’s the real issue: an algorithm can deny your claim in milliseconds, but proving it was unfair might take years of litigation. By the time regulators gather “enough data,” millions of decisions have already been made.
How AI Transparency (or Lack of It) Hits Your Wallet
Let’s get specific about what this regulatory gap means for you:
Premium pricing opacity: Your car insurance uses AI to analyze 200+ data points—driving habits, credit score, even your social media (in some states). But you’ll never see which factors the algorithm weighted most heavily. One insurer might prioritize your commute distance, another your job title. Without disclosure standards, comparison shopping becomes guesswork.
Claims denial without explanation: Health insurers increasingly use AI to flag claims for “medical necessity” review. If the AI decides your MRI wasn’t justified, you’ll get a denial letter citing policy language—not the actual algorithm logic. In states without AI disclosure rules, you can’t even request that information.
Coverage availability blind spots: Some life insurers now use AI to screen applicants before assigning them to underwriters. If the AI filters you out based on correlations in the data (say, your job industry correlates with health risks), you might get rejected without a human ever reviewing your application.
The kicker? In 32 states, you have zero legal right to know if AI made these decisions about you.
What the Proposed NAIC Evaluation Tool Could Change
The tool regulators are debating would create a standardized framework for assessing AI risks in insurance. Think of it as a checklist:
- Data quality verification—does the training data reflect demographic diversity?
- Bias testing protocols—can the insurer prove the AI doesn’t discriminate?
- Human oversight checkpoints—when does an algorithm decision get reviewed by a person?
- Consumer redress pathways—how do policyholders challenge AI-driven outcomes?
Sounds great, right?
The opposition argues this amounts to regulating the technology instead of the outcome. If an AI produces fair results, do regulators need to audit its code? Plus, the evaluation tool would require state insurance departments to develop AI expertise they don’t currently have.
Both sides have valid points. But while regulators debate methodology, your premiums and coverage decisions keep getting made by algorithms with minimal oversight.
Should You Worry About AI in Your Insurance Right Now?
Depends on where you live and what type of insurance you carry.
High concern states (32 without AI guidance): If your state hasn’t adopted the NAIC bulletin or created its own rules, you’re in regulatory limbo. Insurers can use AI with only the loosest interpretation of decades-old laws.
Medium concern states (24 with NAIC bulletin): Better than nothing, but the bulletin lacks teeth. It’s more “best practices guide” than enforceable standard.
Lower concern states (CO, NY, CA, TX with specific rules): You have more protection, but even these states are writing the rules in real-time as AI evolves.
Here’s what you can do regardless:
- Request claim explanations in writing. If your claim gets denied, ask for specific policy provisions that apply. If the letter mentions “system review” or “automated processing,” push for human reconsideration.
- File complaints with your state insurance department. Even without AI-specific rules, pattern complaints about unfair algorithmic decisions could trigger investigations.
- Check if your state has an AI insurance task force. Many states formed working groups in 2024-2025. Public comment periods give you a voice in the rules being written.
- Compare insurers on transparency, not just price. Some carriers voluntarily disclose more about their AI use than others. That’s worth considering during open enrollment.
The 2026 AI Regulation Showdown
Industry experts expect 2026 to be the year AI insurance regulation either gets serious or collapses into state-by-state chaos. Why?
First, New York’s algorithmic bias audit requirements fully kick in by mid-2026. If they uncover significant problems (or don’t), that evidence will shape the national debate.
Second, the NAIC faces pressure to finalize its evaluation tool before more states create incompatible systems. Insurers operating in multiple states are already juggling four different AI compliance regimes—scaling that to 15 or 20 becomes operationally impossible.
Third, federal agencies might step in. The Federal Trade Commission has warned it will enforce existing consumer protection laws against unfair AI practices. If state regulators don’t act, federal intervention becomes more likely.
For consumers, this means the next 12-18 months will determine whether you get meaningful AI transparency or just more bureaucratic shuffling.
Frequently Asked Questions
Can I find out if AI was used to deny my insurance claim?
In most states, no—you don’t have a legal right to that information. Colorado, New York, and proposed California rules give you more disclosure rights, but in the 32 states without AI-specific guidance, insurers aren’t required to tell you if algorithms made the decision. Your best move: request a written explanation citing specific policy provisions, then ask your state insurance department if they have an AI complaint process.
What’s the difference between the NAIC bulletin and state laws on AI?
The NAIC bulletin is a model guideline that states can adopt (or ignore). It has no enforcement power on its own. State laws like Colorado’s annual AI impact assessments or Texas’s $10,000 human review threshold are legally binding regulations with penalties for violations. Think of the NAIC bulletin as a suggestion; state laws are requirements.
Will AI insurance regulations increase my premiums?
Possibly short-term, unlikely long-term. Compliance costs for AI audits, bias testing, and documentation could add 2-3% to operational expenses initially. But if regulations force fairer pricing and reduce discriminatory denials, competition should eventually lower premiums for consumers who were being overcharged. The bigger risk is no regulation—opaque AI pricing makes it impossible to shop effectively.
Which states are most likely to pass AI insurance laws next?
Watch Illinois, Massachusetts, and Washington—all three formed AI insurance task forces in 2024-2025. Florida and Pennsylvania are also considering requirements, though their focus is more on cybersecurity than algorithmic fairness. States with existing comprehensive data privacy laws (like Virginia and Connecticut) might fold AI insurance rules into broader consumer protection updates.
How can I tell if my insurer is using AI responsibly?
Look for voluntary disclosures in their privacy policy or underwriting guidelines. Some insurers publish AI ethics frameworks or participate in industry fairness initiatives like the Insurance Information Institute’s responsible AI programs. Red flags: vague language about “advanced analytics,” no mention of human oversight, or refusal to explain algorithmic decisions when you ask. Also check your state insurance department website—some maintain lists of insurers that have committed to AI transparency standards.
Bottom Line: Transparency Now or Chaos Later
The NAIC’s division on AI regulation isn’t just bureaucratic infighting. It’s a fundamental disagreement about whether consumers deserve to know how algorithms affect their financial lives.
Right now, only 24 jurisdictions have adopted even basic AI guidelines, and only four states have enforceable rules. That leaves over 60% of Americans in a regulatory gray zone where AI makes insurance decisions with minimal oversight.
The proposed evaluation tool could standardize protections nationwide—or it could die in committee while insurers deploy ever-more-sophisticated algorithms. Either way, 2026 will reveal whether regulators can keep pace with technology or if consumers will need federal intervention to get basic AI transparency.
Until then? Document everything, demand written explanations, and file complaints when AI-driven decisions seem unfair. You might not have strong legal protections yet, but your voice in the regulatory process matters more than you think.