Texas TDI Bulletin B-0003-26 Sets AI Governance Expectations for Regulated Entities

Texas Commissioner’s Bulletin B-0003-26, issued June 12, 2026, tells all regulated entities that decisions or actions impacting consumers and made or supported by advanced analytical and computational technologies, artificial intelligence among them, must comply with all applicable insurance laws and regulations. The wording is deliberately wider than AI, so a scoring routine that nobody in the shop calls a model is inside it. The bulletin extends those expectations to any third party working with a regulated entity.

The Texas Department of Insurance is not waiting for a new AI-specific statute. Instead, it is applying existing unfair trade practice, claims settlement, and anti-discrimination rules to AI systems. The bulletin puts it this way: “If a regulated entity uses AI to make a consequential decision, TDI expects a person to review and agree with all decisions before action is taken.”

Human-in-the-loop review stops being a design preference here and becomes a stated regulatory expectation, but the bulletin attaches no recordkeeping obligation to it. What it asks instead is that a regulated entity using AI be able to furnish its procedures and protections when TDI requests them. The bulletin also points to the NAIC Principles on Artificial Intelligence and the Texas Department of Information Resources AI Code of Ethics as reference frameworks for responsible AI use.

The third-party extension is particularly important. A carrier that buys AI models, scoring tools, or automated decision systems does not move the expectations off its own books; they follow the work to whoever performs it. How a carrier makes that hold, whether through contract terms, diligence, or audit rights, is left to the carrier. The bulletin names no clause a vendor agreement has to carry.

One of the statutes the bulletin lists is not a general standard at all. Chapter 4201, concerning utilization review agents, “prohibits using AI to make an adverse determination,” and TDI sets it down in the same list as the unfair trade practice and anti-discrimination chapters. Texas is treating AI governance as an extension of existing consumer protection and market conduct rules, and Chapter 4201 shows what that looks like in practice. The prohibition arrived in SB 815 during the 2025 session and reaches health benefit plans delivered, issued, or renewed on or after January 1, 2026, so it was on the books months before the bulletin issued. The bulletin’s work is to say that it binds a model the same way it binds a person.

The bulletin names the headings an inquiry will follow instead of prescribing specific practices or documentation requirements: “governance frameworks, risk management, data and privacy protections, and internal controls.” A separate sentence encourages carriers to develop verification and testing methods for identifying errors and bias, which sits outside that list of headings. Chapter 401 supplies the teeth, obliging a regulated entity to maintain and produce all books and papers relating to its business on request. What an inquiry reaches, then, is the governance program a carrier already keeps, in whatever form it already keeps it.