AI in Texas Health Care

Short Answer

Texas allows health care practitioners to use AI for diagnostic purposes subject to Health and Safety Code 183.005. The use must stay within the practitioner’s licensed scope and other law, and the practitioner must review AI-created records under the applicable medical-record standards. Diagnostic use requires patient disclosure. TRAIGA separately requires disclosure when AI is used in relation to a health care service or treatment, by the date that service or treatment is first provided, with an emergency timing exception. Insurance Code 4201.156 limits automated adverse utilization-review determinations. HIPAA, device law and nondiscrimination obligations add separate duties. The Attorney General’s Pieces Technologies assurance illustrates consumer-protection scrutiny of advertised AI accuracy; it does not decide whether every clinical AI product is lawful.

Which Laws Apply

Texas AI-specific: Health and Safety Code chapter 183 (SB 1188); Business and Commerce Code 552.051(a) and (f); Insurance Code 4201.156 (SB 815).

Generally applicable Texas law: Medical Practice Act and Texas Medical Board standards; Health and Safety Code chapter 181 (medical records privacy); DTPA.

Federal: HIPAA Privacy and Security Rules; FDA device law; Section 1557 of the Affordable Care Act and 45 C.F.R. 92.210.

Diagnosis and Records (SB 1188)

Health and Safety Code 183.005 permits a practitioner to use AI for diagnostic purposes, including recommending a treatment course based on a patient’s record. The practitioner must act within the licensed scope, the use must not be otherwise restricted or prohibited, and the practitioner must review all records created with AI consistently with Texas Medical Board medical-record standards. The diagnostic use must be disclosed to the patient.

Health and Safety Code § 183.002 requires covered electronic health records to be physically maintained in the United States or a United States territory, including records maintained by a third party or subcontractor. The enactment applies this requirement to records stored on or after January 1, 2026. Chapter 183 also restricts access and requires the biological-sex information specified in Health and Safety Code 183.007 to be included in certain treatment algorithms and decision-support tools. Sections Health and Safety Code 183.010 and .011 provide AG and licensing remedies. A cloud contract should identify actual storage and subcontractor locations, not only the vendor’s headquarters.

Disclosure (TRAIGA)

TRAIGA defines “health care services” as services related to human health, or the diagnosis, prevention or treatment of a human disease or impairment, provided by a licensed, registered or certified individual (Business and Commerce Code 552.051(a)). If AI is used in relation to a health care service or treatment, the provider must disclose that to the recipient or the recipient’s personal representative no later than the date the service or treatment is first provided, or as soon as reasonably possible in an emergency (Business and Commerce Code 552.051(f)). Ambient documentation tools and AI-drafted patient messages are uses a provider should consider under this section.

Coverage Decisions (SB 815)

A utilization review agent may not use an automated decision system to make an adverse determination, the Commissioner of Insurance may audit the agent’s use of such systems, and administrative support and fraud detection are excepted. Notices of adverse determinations must describe the screening criteria and review procedures used. See AI in Insurance and Financial Services.

Accuracy Claims

On September 18, 2024, the Attorney General announced an assurance of voluntary compliance with Dallas-based Pieces Technologies. The AG questioned substantiation of advertised hallucination rates. The filed assurance contains disclosures about metrics, known limitations and harms, and restrictions on unsupported advertising, with covered commitments running through 2029. It is a negotiated resolution of an investigation, not a trial finding. Its practical lesson is to ask what the accuracy measure counts, what data supports it and whether the measure matches the intended clinical task.

Federal Overlay

An AI vendor handling protected health information for a HIPAA covered entity can be a business associate, requiring a compliant agreement and safeguards. FDA’s Clinical Decision Support Software guidance was issued January 6, 2026 and reissued January 29, 2026. Device status depends on the statutory criteria and the software’s function; a marketing label does not settle it. Section 1557 and 45 CFR 92.210 address discrimination in patient care decision-support tools. HHS’s June 2, 2026 Federal Register notice describes partial vacatur of specified gender-identity provisions of the 2024 rule, not blanket withdrawal of all clinical-tool nondiscrimination duties. Covered entities should match the specific requirement and protected ground to the operative rule and court relief.

Illustrative Example (Hypothetical)

Hypothetical: a Fort Worth HIPAA-covered clinic adopts an ambient AI scribe that handles protected health information and suggests diagnoses. A business-associate relationship requires the applicable agreement and safeguards. TRAIGA requires the disclosure by the date the service or treatment is first provided, with its emergency exception; diagnostic use adds Health and Safety Code 183.005. Practitioners need a process to review all AI-created records. Recording the visit raises separate consent questions.

What Is Unsettled

A documentation tool may fall within TRAIGA’s health-service disclosure even where its use is not diagnostic under Health and Safety Code 183.005. Whether a particular feature supplies a diagnosis or treatment recommendation depends on its function. Medical-record review remains a practitioner responsibility; the TMB’s published medical-record standards are the starting point. A combined notice can address both statutes if it gives the required information by the earlier applicable deadline.