Skip to content
Georgia Commons
Menu

Senate · Passed · 2025-2026 Regular Session

SB 444: Private Review Agents; certain decisions with regard to the provision of insurance coverage for healthcare services shall not be based solely on artificial intelligence systems; provide

Last action May 5, 2026 · Act 411

Senate Bill 444 would require that health insurance companies in Georgia have a licensed medical reviewer, not just artificial intelligence, sign off before denying coverage for a healthcare service.

Read the full bill text

These buttons carry the bill's own text, not the summaries below. Copy for LLM, View as markdown, and Send to AI use the Markdown version: the text as filed, then the summaries under a heading that names them as ours. View raw is the text alone.

The summaries below were written by an AI model (claude-sonnet-5) from the text of the bill and are not part of it. Quote the text, not the summary. The stored text is the Enrolled version, the latest the General Assembly's website lists.

In plain language

Georgia insurance companies use private review agents and utilization review entities to decide whether to approve or deny coverage for medical treatments. This bill adds a new section to Georgia's insurance law (O.C.G.A. Chapter 46 of Title 33) addressing how artificial intelligence can be used in that process. The bill allows insurers to use artificial intelligence systems, artificial intelligence, or other software tools to automate tasks and help with decision-making, as long as those tools are part of an approved utilization review plan. However, it prohibits AI from issuing a final denial of coverage (called an adverse determination) to a patient on its own. A human private review agent or utilization review entity, including a clinical peer, must be involved before a denial is finalized, and AI cannot override that person's judgment. The law would take effect January 1, 2027.

What the bill does

  • Adds a new Code section (33-46-7.1) letting insurers use AI systems for utilization review, but only within an approved review plan.
  • Bars AI systems from issuing an adverse determination (a coverage denial) to a patient without a human private review agent or utilization review entity involved.
  • Requires a clinical peer, a medical professional with relevant expertise, to participate before a denial is finalized.
  • States that AI tools cannot override or supersede the judgment of the clinical peer reviewing the case.
  • Defines 'artificial intelligence' and 'artificial intelligence system' for purposes of this insurance law.
  • Sets the law to take effect January 1, 2027.

Who it affects

Health insurance companies and their private review agents and utilization review entities in Georgia, the clinical peers (licensed medical professionals) who review coverage decisions, and patients whose insurance claims go through utilization review for healthcare services.

Why it matters

Patients seeking insurance approval for medical care would be guaranteed that a human medical reviewer, not just an automated algorithm, has a role in any decision to deny coverage. This could affect how quickly and how coverage denials are issued by Georgia insurers.

Key provisions

  • Section 1 adds Code Section 33-46-7.1 defining 'artificial intelligence' and 'artificial intelligence system' for insurance review purposes.
  • Subsection (b) permits private review agents and utilization review entities to use AI tools if they are part of a compliant utilization review plan approved under existing standards and Commissioner rules.
  • Subsection (c) allows AI to automate tasks and assist decision-making but bars it from issuing an adverse determination without a human private review agent or utilization review entity and a participating clinical peer.
  • Subsection (c) also states AI tools cannot supersede the judgment of the clinical peer involved in the review.
  • Section 2 sets the effective date as January 1, 2027.
  • Section 3 repeals conflicting laws.

From the bill

“such systems shall not issue an adverse determination to a patient until a natural person qualifying as a private review agent or a utilization review entity conducts a utilization review in which a clinical peer participates.”

This requires a human medical reviewer to be involved before AI can be used to deny a patient's coverage.

“In no event shall artificial intelligence systems, artificial intelligence, or other software tools supersede the judgment of such clinical peer.”

This ensures AI recommendations cannot override the decision of the human clinical reviewer.

Discussed in meetings (16)

The House and Senate meetings that took this bill up, from their machine transcripts. Each account was written by Georgia Commons with AI and is not part of the record; the video is what was said.

  1. Senate Insurance and Labor Committee, February 5, 2026

    From 1:00:11 · Watch on the General Assembly's Vimeo channel

    Summary by Georgia Commons: The sponsor said SB 444 would allow artificial intelligence to help process requests for advance approval of medical care, but prohibit its use as the sole basis for an unfavorable coverage decision. The sponsor said a clinical peer would need to review the case before a denial, and that reviewer’s decision would prevail. The sponsor described the goal as speeding a time-consuming process while ensuring the full medical record is considered. One member asked whether patients should be notified when artificial intelligence is used. The sponsor offered to consider the issue but questioned where notification would fit, explaining that providers generally communicate with insurers and that a denial would not be made solely by artificial intelligence. The same member raised concerns about discriminatory algorithms and pricing. The sponsor responded that the bill concerns medical necessity, not pricing, and emphasized human review before a final denial. The chair described that review as a patient protection. No amendment was offered. The committee approved the bill 6–1, and the chair announced that it would go to Rules.

  2. Senate Rules Committee, February 9, 2026

    From 0:24:40 · Watch on the General Assembly's Vimeo channel

    Summary by Georgia Commons: The sponsor described the bill as a consumer protection measure concerning prior authorization for health care. The sponsor said that when AI is used, a human clinical peer would have to participate before an adverse decision is made. No questions or vote were shown.

  3. Senate Rules Committee, February 10, 2026

    From 0:08:05 · Watch on the General Assembly's Vimeo channel

    Summary by Georgia Commons: The sponsor described SB 444 as a patient-protection measure to ensure human involvement when artificial intelligence is used in prior authorization decisions. No substantive questions followed. Later, the sponsor selected the bill to move forward, and the committee unanimously approved it together with two other bills.

  4. Senate floor session, February 11, 2026 (legislative day 17)

    From 2:06:21 · Watch on the General Assembly's Vimeo channel · Confirmed by the bill's status history (Senate Third Read)

    Summary by Georgia Commons: The sponsor said SB 444 would protect patients by requiring a human clinical peer to participate before an adverse insurance coverage determination is made. The sponsor explained that insurers often require approval before treatment and said the bill recognizes artificial intelligence's role while preventing certain decisions from being based solely on AI or other software. The Insurance and Labor Committee recommended passage. No questions or opposing speeches followed, and the Senate passed the bill 49-0.

  5. House Technology and Infrastructure Innovation Committee, March 5, 2026

    From 0:27:00 · Watch on the General Assembly's Vimeo channel

    Summary by Georgia Commons: The sponsor said the bill would allow artificial intelligence in insurance authorization reviews but prohibit it from being the sole basis for an adverse coverage decision or overriding a clinical peer’s judgment. The sponsor said human review can account for circumstances missing from medical records, cited Stanford research raising concerns about algorithms and oversight, and cited an American Medical Association paper reporting higher denial rates with AI. Members questioned whether the bill’s definitions should match language developed elsewhere in Georgia law and accommodate future technology. The chair proposed working on revised definitions before a second hearing. The sponsor welcomed changes, explaining that the current wording came from the National Institute of Standards and Technology. Another member noted that definitions can appropriately differ by legal context, including insurance. A Medical Association of Georgia witness supported the bill, arguing that qualified human review would help prevent automated errors and bias and preserve accountability for decisions affecting care. A member also suggested discussing language distinguishing human intelligence from AI. The chair planned a small-group discussion with the sponsor and interested members before the next hearing. No amendment was adopted and no vote was taken.

  6. House Technology and Infrastructure Innovation Committee, March 11, 2026

    From 0:13:16 · Watch on the General Assembly's Vimeo channel

    Summary by Georgia Commons: The sponsor said the bill would allow AI in medical prior authorization but prohibit using it as the sole basis for denying care, preserving review by a qualified clinical peer. Asked about potential loss of federal broadband funding under an executive order, the sponsor said the current administration had raised no objection to this narrowly focused bill. Members questioned whether private review agent meant an AI system; the sponsor and committee attorney explained that existing law defines the term as a person or entity performing insurance utilization review. Asked about disclosure to patients, the sponsor said disclosure should occur through general intake paperwork but was not specifically required by this bill. The chair proposed expedited review after denial, while the sponsor said review needed to occur before denial. Another member repeatedly sought assurance that a human must thoroughly review a potential denial. The sponsor said existing law required that process, but members requested more explicit language. The chair also suggested replacing the AI definition with language developed for HB 171 for consistency. No amendment was adopted. The chair asked members and the sponsor to refine the bill before Tuesday, and no vote occurred.

  7. House Technology and Infrastructure Innovation Committee, March 17, 2026

    From 0:33:07 · Watch on the General Assembly's Vimeo channel

    Summary by Georgia Commons: After initially postponing consideration while presenters were unavailable, the committee heard a substitute from a committee member presenting on the sponsor’s behalf. The presenter said the substitute clarified that artificial intelligence and other software could not override a clinical peer’s judgment in insurance coverage reviews, keeping a human involved in the decision. The presenter also said the substitute used the House’s definition of an artificial intelligence system and that the sponsor supported the changes. In response to the chair’s question, the presenter confirmed that language requiring a clinical peer’s participation was newly added. With no further questions or discussion, the committee approved the substitute by voice vote and sent it to Rules. No vote count was stated.

  8. House Rules Committee, March 19, 2026

    From 0:38:46 · Watch on the General Assembly's Vimeo channel

    Summary by Georgia Commons: The sponsor said the bill would require human involvement before an insurer makes an adverse prior-authorization decision using artificial intelligence. One committee member emphasized having a person explain a denial to patients or providers. The sponsor said existing protections require review by a clinical peer and that AI should not override that professional judgment. Another member asked whether human review would delay approvals. The sponsor said AI approvals could proceed, while potential denials needed a thorough review of the medical record. Asked whether a denial should come from a provider licensed in Georgia, the sponsor agreed and cited the existing definition of a clinical peer. The committee later placed the bill on the Legislative Day 34 calendar without voiced opposition.

  9. House floor session, March 19, 2026 (legislative day 34)

    From 2:08:37 · Watch on the General Assembly's Vimeo channel · Confirmed by the 2026 House Journal, p. 2875

    Summary by Georgia Commons: The clerk described SB 444 as prohibiting certain health-insurance coverage decisions based solely on artificial intelligence. The House sponsor said collecting data to support decisions was reasonable, but a human should make a decision denying treatment. There were no questions. The House adopted the committee substitute without objection and passed the bill, 166 to 0.

  10. Senate floor session, March 25, 2026 (legislative day 37)

    From 7:29:51 · Watch on the General Assembly's Vimeo channel · Confirmed by the bill's status history (Senate Agreed House Amend or Sub)

    Summary by Georgia Commons: The sponsor said SB 444 would require human involvement before an insurer denied medical care through prior authorization, rather than allowing denial solely through artificial intelligence. The sponsor said AI could still assist with coverage verification and approvals, and illustrated concerns by describing an AI-generated wrist diagram with incorrectly named bones. The sponsor said House changes revised the AI definition and clarified that AI judgment could not override a clinical peer's judgment. No questions or opposition followed. The Senate agreed to the House substitute 47 to 0.

Status timeline

  1. 2027-01-01Effective Date
  2. 2026-05-05Act 411
  3. 2026-05-05Senate Date Signed by Governor (Senate)
  4. 2026-04-10Senate Sent to Governor (Senate)
  5. 2026-03-25Senate Agreed House Amend or Sub (Senate)
  6. 2026-03-19House Passed/Adopted By Substitute (House)
  7. 2026-03-19House Third Readers (House)
  8. 2026-03-18House Committee Favorably Reported By Substitute (House)
Show full history (16 actions)
  1. 2026-02-17House Second Readers (House)
  2. 2026-02-12House First Readers (House)
  3. 2026-02-11Senate Passed/Adopted (Senate)
  4. 2026-02-11Senate Third Read (Senate)
  5. 2026-02-09Senate Read Second Time (Senate)
  6. 2026-02-06Senate Committee Favorably Reported (Senate)
  7. 2026-02-03Senate Read and Referred (Senate)
  8. 2026-02-02Senate Hopper (Senate)

Sponsors

  • Kay Kirkpatrick (R, SD-32)Primary sponsor
  • Ben Watson (R, SD-1)
  • Mike Hodges (R, SD-3)
  • Ed Harbison (D, SD-15)
  • Larry Walker, III (R, SD-20)
  • Lee Hawkins (R, HD-27)

Votes

  1. PassedSenate voteFebruary 11, 2026

    49 yea, 0 nay (0 not voting, 5 excused)

    PASSAGE: Senate Vote #540

  2. PassedHouse voteMarch 19, 2026

    166 yea, 0 nay (3 not voting, 7 excused)

    PASSAGE: House Vote #729

  3. PassedSenate voteMarch 25, 2026

    47 yea, 0 nay (2 not voting, 5 excused)

    AGREE TO HOUSE SUBSTITUTE: Senate Vote #821

Topics

  • health insurance
  • artificial intelligence regulation
  • insurance coverage denials
  • utilization review
  • healthcare policy

Ask about this bill

Answers come from this document. Not legal advice.

Machine-readable https://georgiacommons.org/bills/2025-2026/sb444.md · https://georgiacommons.org/bills/index.md · MCP https://mcp.georgiacommons.org/mcp

SB444: Private Review Agents; certain decisions with regard to the provision of insurance coverage for healthcare services shall not be based solely on artificial intelligence systems; provide | Georgia Commons