SB 1822 introduces a new section—4201.156—into the Texas Insurance Code to regulate the use of artificial intelligence (AI)-based algorithms by health insurance issuers and Health Maintenance Organizations (HMOs) in the context of utilization review (UR). Utilization review refers to the process by which insurers evaluate the necessity, efficiency, and appropriateness of healthcare services provided to patients. This legislation is a proactive response to the increasing role of AI in healthcare decision-making, aiming to ensure transparency, accountability, and fairness in its application.
The bill requires health insurers and HMOs to publicly disclose whether they or their utilization review agents employ AI algorithms in the review process. This disclosure must be made available both online and in writing to policyholders, enrollees, and healthcare providers who interact with the insurance issuer. In addition to transparency, the bill mandates safeguards against bias, requiring that any AI algorithm used in UR must be trained on data that minimizes risks of discrimination based on characteristics such as race, gender, age, religion, disability, and other protected classes. Furthermore, the algorithms must conform to evidence-based clinical guidelines to ensure that medical determinations are grounded in widely accepted healthcare standards.
Annually, by December 31, insurance entities subject to the law must file an AI compliance statement with the Texas Department of Insurance. This report must include a summary of the AI’s function, a decision-making logic tree, a description of the datasets used for training the algorithm, and a formal attestation that the system complies with bias-reduction and evidence-based criteria. These provisions aim to provide oversight and public accountability in the growing field of AI-enabled healthcare, balancing innovation with ethical and clinical standards.
The Committee Substitute for SB 1822 introduces several key refinements to the originally filed version, reflecting a shift toward more practical implementation and better protection of proprietary information. While both versions require health insurance issuers and HMOs to disclose whether they use artificial intelligence (AI) in utilization review processes, the substitute strengthens this requirement by explicitly mandating the disclosure be placed on a publicly accessible part of the insurer’s website and shared directly with enrollees and providers.
One of the most significant differences lies in the annual reporting requirements. The originally filed version required insurers or their utilization review agents to submit the actual AI algorithm and its training datasets to the Texas Department of Insurance. This raised potential concerns over the exposure of proprietary systems and trade secrets. In contrast, the committee substitute replaces this with a less invasive “artificial intelligence compliance statement.” Instead of handing over algorithms, insurers must provide summaries of the AI system’s function, a logic or decision tree, a description of the training data sources, and an attestation that these tools comply with anti-bias requirements and evidence-based clinical guidelines.
This revision suggests a clear intent to balance transparency and accountability with the need to protect innovation and business confidentiality. It also reflects responsiveness to industry concerns while maintaining regulatory oversight. By replacing hard data submission with structured summaries and attestations, the substitute provides a clearer regulatory pathway that is both enforceable and less burdensome. These changes indicate a legislative refinement aimed at encouraging compliance without stifling the development and use of AI tools in healthcare.