Senate Commerce Committee
Aaron Jones 271-2609
HB 1406, relative to health carrier recordkeeping requirements in utilization review, including specifications regarding the use of artificial intelligence.
Hearing Date: April 7, 2026
Time Opened: 11:11 a.m. Time Closed: 11:42 a.m.
Members of the Committee Present: Senators Innis, Birdsell and Fenton
Members of the Committee Absent : Senators Ricciardi, Murphy and Reardon
Bill Analysis: This bill:
I. Requires health carriers under the managed care law to maintain written records relative to their use of artifical intelligence and protocols for ensuring human review of certain determinations.
II. Requires health carriers to make any adverse determinations relative to a covered individual and maintain records related to an adverse determination including whether artificial intelligence was used.
III. Establishes health carriers as responsible for the activities of entities which they have contracted with, relative to such requirements.
Sponsors:
Rep. Gregg | Rep. Nagel | Rep. Grossman |
Rep. LaMontagne | Rep. M. Pearson | Rep. Tellez |
Rep. Salvi | Rep. Manohar | Rep. N. Murphy |
Sen. Long | Sen. Birdsell | Sen. Prentiss |
________________________________________________________________________________
Who supports the bill: 135 individuals were in support. Full sign in sheets are available upon request by contacting the Legislative Aide, Aaron Jones (aaron.jones@gc.nh.gov).
Who opposes the bill: 5 individuals were in opposition. Full sign in sheets are available upon request by contacting the Legislative Aide, Aaron Jones (aaron.jones@gc.nh.gov).
Who is neutral on the bill: 2 individuals were neutral. Full sign in sheets are available upon request by contacting the Legislative Aide, Aaron Jones (aaron.jones@gc.nh.gov).
Summary of testimony presented in support:
Representative Alicia Gregg
- Nationally, major health insurers have faced lawsuits because they are using AI systems to review complex insurance claims in as little as 1.2 seconds.
- When denials made by AI are appealed, 97 percent of them are overturned. Representative Gregg said this technology is not efficient; instead, it is automatically denying care to patients and requiring them to go to court to get medical coverage.
- AI systems can be shaped to produce specific outcomes. If an insurer would like to reduce costs, they can train AI to find reasons to deny claims.
- AI systems also struggle with carefully reviewing patients who have complex medical histories involving multiple diagnoses.
- This bill would build on existing protections by adding three specific requirements for AI.
- First, a qualified health care provider must make a final decision on any denial or payment reduction when AI is used in claims processing. While AI can assist in flagging items for review, a real human with medical training must make the final call.
- Second, insurers must tell patients and providers when AI has been used in a decision.
- Third, insurers must keep detailed records about how they are using AI to allow the Insurance Department to audit what is happening.
- When AI automatically reduces payment for services, it takes money away from providers who delivered the necessary care. It also creates a chilling effect for providers to code conservatively to avoid down coding. As a result, providers are not accurately documenting the complexity of the care they are providing.
- Instead of creating a new framework, Representative Gregg worked with the Department to add requirements to the existing law.
- Similar protections have been enacted in Arizona, California, Maryland, and Texas, and they are being considered in 19 other states.
- This bill would not prevent AI from being used for administrative tasks or to prevent fraud. When it comes to denying or reducing health care claims, however, providers should make the final call.
Adam Thompson
- Mr. Thompson was concerned about the possibility of an algorithm down coding or changing a provider or care team’s recommendations to prioritize the bottom line of a carrier rather than the health of a patient.
- This bill would establish guardrails within the existing utilization review framework, require transparency in how AI is used, ensure documentation of its role in decision-making, and require adverse determinations to be made by a qualified health care provider.
- Decisions that deny care, reduce payments, or change provider coding should not be made solely by automated systems.
- Clinical decisions are not entirely data driven, they rely on professional judgment, patient history, relationships between patients and their care teams, and context that cannot be captured by an algorithm.
- For individuals with disabilities, chronic conditions, or neurodivergent profiles, Mr. Thompson said care was rarely simple or predictable. Often, he said individuals with atypical medical profiles are first to experience friction within rigid systems.
Andrew Horn
- Mr. Horn said modern AI is built on random number generation, which includes complex math that even its creators do not fully understand.
- Mr. Horn said they were not asking machines to be silent, they were only asking for human beings to have the final word. He said a medical decision without sign off from a provider was like a ship without a captain.
- This bill would provide transparency to the decision-making process thereby ensuring accountability.
- Even if a decision is delegated to a contractor, Mr. Horn said an insurance company must be responsible for their decisions.
- This bill would protect patients, providers, and insurance companies by keeping medical decisions in human hands.
Summary of testimony presented in opposition:
Sabrina Dunlap, Senior Director of Government Relations, Anthem Blue Cross Blue Shield
- Ms. Dunlap said they do not use AI to deny claims; instead, it is used to catch patterns of fraud, waste, and abuse that cost the health care system billions each year.
- Ms. Dunlap said any decisions involving clinical judgment are signed off on by a human.
- Ms. Dunlap said it is difficult to legislate this dynamic and fast-moving area of technology.
- This bill could have adverse consequences. It is also unnecessary because the Department already has regulatory authority over this area.
- This bill would not regulate the use of AI by providers to find codes that maximize their revenue.
- Senator Fenton asked if AI determined what humans took a closer look at.
- Ms. Dunlap said AI was used to make the process faster. If a claim has been denied, an individual will review it and a decision will be made.
- Senator Fenton said he feared that while AI could be helpful, things could fall through the cracks if it is used to flag certain things. He asked if AI influenced whether something was flagged.
- Ms. Dunlap said AI does not change the process of how claims that are not approved are reviewed; instead, it is used to help to move things along. AI does not change instances where something will be denied.
Cam Lapine, on behalf of Cigna
- Page 13 of INS 2002.07 addresses the use of AI, and Insurance Department Bulletin 24-011-AB provides an overarching AI bulletin that applies to all insurers.
- Mr. Lapine said this bill went beyond a “belt and suspenders” approach because certain provisions went beyond what was in existing regulations. For example, on Line 13, it states, “all adverse determinations, including payment reductions and down coding…”. Adverse determinations are defined in RSA 420-J:3, and they involve the review of clinical material for the determination of medical necessity. Payment reductions and down coding do not meet the definition of adverse determinations. If they are included in this provision, he said it is an expansion of existing regulation as well as poses a conflict with the existing statutory definition.
- Requiring a qualified health care provider to be involved in administrative functions would increase costs and add delays to patient care.
- Senator Fenton asked if requiring documentation and disclosure of its use would not change current practices, especially if AI was not used to deny or influence claims.
- Mr. Lapine said the inclusion of payment reductions and down coding, which are not adverse determinations, would be a change from current regulation. The documentation language would be new by putting it in statute.
- Senator Fenton asked if their processes on denials would not be influenced or changed if the use of AI has been documented.
- Mr. Lapine said he did not believe so.
Paula Rogers, on behalf of America’s Health Insurance Plans
- When the bulletin was issued by the Department, they did not advise the plans prior to its issuance.
- While down coding has a negative connotation, Ms. Rogers said there can be up coding on the provider’s side.
- Codes for medical procedures are an administrative function, and this bill would say that a qualified health care professional would be required.
- If the AI system is well programmed, Ms. Rogers said it would likely look at problematic or irregular trends.
- An appeals process does exist if a provider or a patient feels they have been deprived.
- Ms. Rogers said they would be comfortable if the reference to down coding were removed.
- Ms. Rogers said New Hampshire would be the only state to adopt the provisions on Line 14.
Neutral Information Presented:
Michelle Heaton, Director of Life and Health, and Josh Hillard, New Hampshire Insurance Department
- The Department issued a bulletin that established guardrails on how AI is regulated. They are in the process of updating their INS 2000 and INS 1300 rules related to utilization review and claim processing, respectively.
- Ms. Heaton said they were trying to align with NAIC and their definitions on AI to ensure consistency with other states.
- The amendments made in the House ensured that when AI is used, there is appropriate documentation on how it is used and who is making the decisions.
- Utilization reviews are medically necessary determinations on whether certain care is appropriate. Ms. Heaton said existing laws already require an individual to make those decisions.
- Once a decision has been made, there is an appeal and external review that is conducted by medical professionals. There are already regulations in place on who can make those decisions.
- Senator Fenton asked if there were any gaps in how carriers disclose how they use AI under current law.
- Ms. Heaton said no. Under their regulatory structure, they have broad authority to ask carriers for anything, including their use of AI and its related documentation. It is already difficult to teach consumers how to use their insurance policies let alone the complexities of claim adjudication. They are actively developing tools to monitor consistency among other states.
- Mr. Hillard said the Department has been in communication with NAIC, which has several subcommittees dealing with these issues. It has been a work in progress to allow for innovation, while also safeguarding consumers.
- Senator Fenton asked if there were any gaps.
- Ms. Heaton said not in their current system.
- Senator Birdsell asked if this was a “belt and suspender” type of legislation since insurance companies are not allowed to use AI to deny claims.
- Ms. Heaton said while it is not explicitly stated in statute, it is not allowed under the current structure. The purpose of this legislation would be to explicitly state that if AI is used, documentation must be kept.
AJ
Date Hearing Report completed: April 13, 2026