2025 Texas sessionsRegular and both called sessions have ended. This index covers House and Senate bills, not resolutions. Statuses come from official reports.

2025 · Regular session · Senate bill

SB815

SB815 limits automated denials in health-plan utilization review—the process of checking whether care is medically necessary or appropriate.

Read bill highlights ↓

Official caption: Relating to the use of certain automated systems in, and certain adverse determinations made in connection with, the health benefit claims process.

Signed by governorSource-checked explanation published

Official report lists Senate author(s): Schwertner. This does not identify sponsors or amendment drafters.

Report an error ↗

Bill explanation published · Version and amendment review scope is stated below · Individual amendment records may not be separately annotated.

Bill highlights

Selected points from the published explanation. Follow each point to its source-linked context; conditions and limits may matter.

SB815 limits automated denials in health-plan utilization review—the process of checking whether care is medically necessary or appropriate.
The law defines an algorithm as a computerized set of steps, AI as a machine-learning system that infers outputs from inputs, and an automated decision system as an algorithm using data-based analysis to make, suggest or recommend determinations.
The governor signed SB815 on June 20, 2025.
Coverage still follows Insurance Code Chapter 4201; this is not a promise that every Texas insurance denial falls under the new rule.

Read the full explanation ↓

What does it do?

SB815 limits automated denials in health-plan utilization review—the process of checking whether care is medically necessary or appropriate. A review agent may not use an automated decision system to make an adverse determination, wholly or partly. The definition also covers findings that care is experimental or investigational. This is not a ban on all insurance AI: the law expressly permits administrative support and fraud detection. Final law, sections 1–2.

The law defines an algorithm as a computerized set of steps, AI as a machine-learning system that infers outputs from inputs, and an automated decision system as an algorithm using data-based analysis to make, suggest or recommend determinations. The insurance commissioner may audit or inspect a review agent’s use of automated systems at any time. Denial notices must give both a description of and the source of the screening criteria and review procedures used—not merely one or the other. Existing requirements for reasons, clinical basis and appeal information remain. Definitions, oversight and notice requirements.

The governor signed SB815 on June 20, 2025. It took effect September 1, 2025, but its changes apply only to review for health benefit plans delivered, issued for delivery or renewed on or after January 1, 2026. Earlier plans remain under the prior law for this purpose. Official history; sections 4–5.

Coverage still follows Insurance Code Chapter 4201; this is not a promise that every Texas insurance denial falls under the new rule. The chapter includes health maintenance organizations (HMOs), expressly including Medicaid managed-care HMOs, while retaining exceptions for specified federal contracts and state programs. It does not apply to the terms or benefits of employee welfare benefit plans defined by the federal Employee Retirement Income Security Act (ERISA). That last exception should not be read here as a blanket statement about every employer-plan review. HMO scope; federal-contract exception; state-program exceptions; employee-plan provision.

What changed along the way?

The introduced bill would have prohibited an AI algorithm from being the sole basis for an adverse determination. The Senate committee substitute broadened the proposed restriction to automated decision systems, proposed AI disclosures in explanations of benefits, and proposed a new requirement for an appropriate Texas-licensed physician, dentist or other health-care provider to make adverse determinations. For physician services, it would have required a physician with an unrestricted Texas license and the same or a similar specialty.

The official amendment index lists one amendment: Schwertner’s Senate second-reading F1, adopted March 26. Its full text removed both explanation-of-benefits disclosure sections and the proposed new decision-maker section. It also removed “support” from the automated-system definition, replaced the broader proposed restriction with the final “make, wholly or partly” rule, and expressly allowed administrative support and fraud detection. These are substantive limits on what became law; the deleted disclosures and new personnel provision were not enacted by SB815.

Comparison of the five indexed versions confirms that the engrossed, House committee and final operative texts match. The final law retains the broader adverse-determination definition and stronger notice language. Removal of the proposed personnel section does not remove existing clinical safeguards: utilization review remains under a Texas-licensed physician’s direction. A separate 2025 law, HB3812, added the restriction against an administrative-medicine-only license; that change is not credited to SB815. Final SB815; physician-direction law and amendment history.

For the full amendment record, open the official amendment index.

Official bill versions

Compare the published text versions in the official text index. A change is not attributed to a specific amendment here without a verified record.

Who could benefit—and who could bear the costs?

Patients whose plans and reviews are covered gain a legal restriction on automated adverse decisions and fuller information about the criteria used. Review agents may need to change decision workflows and notices and be ready for state inspection. These protections do not guarantee approval of a treatment, eliminate all denials or establish a particular reduction in premiums or patient costs. Final requirements and limits.

The April 22 engrossed fiscal note—for text matching the final operative law—anticipated no significant state or local fiscal impact and assumed costs could be absorbed with existing resources. It does not quantify private compliance costs or individual patient savings and is not an appropriation.

The version matters: the February 25 introduced note also expected no significant state or local impact, but the March 13 committee-substitute note said Medicaid client-service effects could not be determined because existing managed-care use of automated systems was unknown. That earlier uncertainty describes the broader proposal, not a quantified cost assigned to the final law.

Official sources

Catalog report retrieved 2026-10-25. Check the official record for corrections or later changes.