AI governance for Texas government & hospital districts
Texas state agencies, local governments, and hospital districts now face a distinct public-sector AI stack — SB 1964, HB 3512, and TRAIGA, with SB 1188 potentially in play for hospital districts. Risk Meridian gives public-sector teams one platform to classify governmental AI systems, document rationales, track training, and produce procurement-, board-, and audit-ready evidence.
No credit card required.
The public-sector AI stack is already in effect
SB 1964 and HB 3512 both took effect September 1, 2025, and TRAIGA comes into force January 1, 2026. For governmental entities, that means AI use, data management, procurement, ethics, DIR interaction, personnel training, and TRAIGA disclosure duties are converging on the same teams at the same time. The agencies that document their governance now — with written rationales, versioned assessments, and a clean training record — are the ones ready when procurement, a board, or an auditor asks. Specific obligations and applicability should be confirmed against each statute and with counsel.
Your obligations, statute by statute
SB 1964 and HB 3512 apply to governmental entities — state agencies and local governments. A hospital district may qualify and may also reach SB 1188. Confirm applicability with counsel; Risk Meridian helps you document across every framework.
SB 1964
Effective Sept 1, 2025
SB 1964 adds Subchapter S to Government Code chapter 2054 and governs how state agencies and local governments procure, develop, deploy, and use AI. Its sharpest duties attach to a Heightened Scrutiny AI system (HSAI) — AI intended to autonomously make, or be a controlling factor in, a consequential decision that materially affects access to a government service, subject to four statutory exclusions (§ 2054.003). Covered entities shall adopt the DIR AI Code of Ethics (§ 2054.702) and DIR minimum standards for HSAI (§ 2054.703), inventory AI and HSAI with a purpose and risk-mitigation evaluation (§§ 2054.068, 2054.0965), conduct confidential HSAI impact assessments (§ 2054.708 — mandatory for state agencies; local governments are directed to consider one under 1 TAC § 219.23(e)(2)), and post the DIR standardized notice (§ 2054.711). Enforcement lands on vendors through a 31-day cure-then-void path that can bar repeat offenders from state contracts (§§ 2054.709–710). DIR's implementing rules were adopted as 1 TAC Chapter 219, effective March 18, 2026 with no transition period — an entity that has not yet designated an AI Risk Officer or adopted the Code of Ethics is currently out of conformance.
HB 3512
Effective Sept 1, 2025
HB 3512 requires annual AI training for public-sector personnel. A state agency must identify each employee who uses a computer for at least 25% of their duties; those employees and every elected or appointed officer must complete a DIR-certified AI training program at least once a year (§§ 2054.5191(a), 2054.5193). Local governments carry the same duty for employees and officials with computer-system access (§ 2054.5191(a-1)). Completion is verified and reported to DIR on its form and is subject to periodic audits, and each state agency certifies compliance in its strategic plan (§ 2056.002(b)(12)). Risk Meridian tracks completion and expiry and produces that reporting.
TRAIGA (HB 149)
In force Jan 1, 2026
TRAIGA is an intent-based prohibition statute enforced exclusively by the Texas Attorney General, with a 60-day cure period and no private right of action. Beyond the general prohibitions, government agencies must disclose AI-to-consumer interactions (§ 552.051(b)) and are barred from AI social scoring (§ 552.053) and certain biometric identification (§ 552.054). TRAIGA starts from a rebuttable presumption of reasonable care (§ 552.105(c)); if the AG investigates, a civil investigative demand can require a system’s purpose, data, outputs, limits, and oversight process (§ 552.103). NIST AI RMF substantial compliance is a named affirmative defense (§ 552.105(e)), and for licensed personnel a licensing agency may add sanctions up to $100,000 after a violation finding and AG recommendation (§ 552.106) — so a documented governance record is how you evidence good faith.
SB 1188 (hospital districts)
Effective Sept 1, 2025
A hospital district is a healthcare provider, so TRAIGA's patient AI-use disclosure duty (§ 552.051(f)) and SB 1188 (record review, EMR-offshoring limits, patient notification) apply to it. Note an important nuance: TRAIGA expressly excludes a hospital district from its definition of “governmental entity” (Bus. & Com. Code § 552.001(3)), so TRAIGA's governmental-only prohibitions do not bind districts under that chapter. Districts may still be reached by SB 1964 and HB 3512 on those statutes' own terms, and SB 1964 gives them a specific break: a district may satisfy the standardized-notice disclosure by including a generalized statement in its patient consent forms that an AI system may be used in the course of treatment (§ 2054.711(c)). Whether and how each applies to your district should be confirmed with counsel — Risk Meridian is built to document across all of them.
How Risk Meridian helps public-sector teams
A government module built for the depth the public sector actually faces — governmental-AI classification, DIR-ready packs, training tracking, and the enterprise rigor procurement and audit expect.
Governmental-AI Classifier
Classify each AI system against the SB 1964 Heightened Scrutiny test: does it autonomously make, or act as a controlling factor in, a consequential decision affecting access to a government service, or does it fall within one of the four statutory exclusions (§ 2054.003)? Every determination produces an attributable, reviewable written rationale. The model is binary, matching the statute: an HSAI determination triggers the extra controls, and every non-HSAI system carries inventory and oversight duties only — no LOW/MODERATE/HIGH tier to argue about.
Versioned Assessments
SB 1964 minimum standards require each HSAI to be assessed and documented for known security risks, performance metrics, and transparency measures before deployment and again at every material change (§ 2054.703). Assessments are versioned, so each pre-deployment and material-change evaluation stays intact as part of the record.
DIR Submission Pack (Local Governments)
Assemble the local-government on-request review for DIR (§ 2054.0965(c)): the AI and HSAI inventory with each system's purpose and risk-mitigation evaluation (§§ 2054.0965, 2054.068), the standardized public notices (§ 2054.711), and any impact assessments your entity has elected to conduct — for a local government the § 219.23(b) impact assessment is advisory, not mandatory (1 TAC § 219.23(e)(2)) — organized from your live records.
IRDR AI Answer Set (State Agencies & Universities)
For state agencies — including institutions of higher education — assemble the information-resources deployment review AI answers (§ 2054.0965(b)(6)–(7)): per-system uses-AI, heightened-scrutiny, and risk-mitigation responses with a purpose evaluation and strategic-plan-support analysis, plus the (b)(7) compliance confirmation built from your code-of-ethics and minimum-standards adoption records (§§ 2054.702(c), 2054.703(c)). Field mappings follow current DIR guidance — verify against DIR's current instrument.
Classification Crosswalk
Many institutional AI policies were drafted in TRAIGA-lineage vocabulary — high-risk, substantial factor, algorithmic discrimination — while SB 1964 and 1 TAC Chapter 219 duties key to heightened scrutiny and controlling factor, with a different exclusion set. The Classification Crosswalk reconciles the two vocabularies for every system in your inventory, with a written reconciliation rationale per system — the artifact an auditor asks for at that seam.
Vendor-Clause Tracking
The adopted rules put two distinct clauses in every AI vendor contract — never merged: the risk-framework clause requiring an AI risk management framework such as NIST's or a comparable standard like ISO/IEC 42001 (1 TAC § 219.24(d)), and the vendor ethical-principles clause binding the vendor to the entity's AI ethical principles and relevant laws (§ 219.11(j)(2)(B) — DIR refused to let vendors substitute their own principles). Track both across your agreements and record which framework each vendor uses — and stay ahead of the vendor enforcement path (§ 2054.709), where an uncured violation runs 31 days to cure, then 31 more after a notice of intent to void, before a contract can be voided.
Three-Track Training Tracker
Texas public-sector AI training is three distinct duties, and no single course satisfies them all: annual DIR-certified AI training for 25%-computer-use employees plus elected and appointed officers and officials (§§ 2054.5191, 2054.5193); all-employee Acceptable Use Policy training (1 TAC § 219.24(b)); and per-HSAI risk training for the employees and contractors who access, use, or manage each heightened-scrutiny system (§ 219.24(c)) — Contractor is a tracked person type. Each is recorded separately, with the annual verification and strategic-plan certification record (§ 2056.002(b)(12)) ready when DIR asks.
Standardized-Notice Generator
Generate the SB 1964 standardized notice for public-facing or consequential-decision AI systems — general information about the system, its data sources, and the measures used to maintain privacy-law and ethics compliance (§ 2054.711), on the DIR form as adopted. For hospital districts, produce the § 2054.711(c) variant — a generalized statement in patient consent forms that an AI system may be used in the course of treatment. Note: whether that consent-form statement also satisfies the separate disclosure duty in DIR's Chapter 219 ethics rule (§ 219.11(g)(2)(C)) is unresolved — DIR declined to confirm it; check with counsel.
Enterprise Rigor & Audit Trail
Multi-tenant RBAC keeps agency data separated by role, a tamper-evident audit log records every action, and PDF artifacts give you portable evidence. Security is Encrypted (in transit and at rest), single sign-on with Google and Microsoft (OIDC), TOTP multi-factor authentication with a server-enforced, organization-level required policy, RBAC, and a tamper-evident audit log. Hosted on AWS in United States regions; customer data is stored and processed in the United States.
Code-of-Ethics Obligations Checklist (§ 219.11)
The widest obligation in the adopted rules — and the one most tooling under-scopes — covers all AI systems the entity procures, develops, deploys, or uses, not just heightened-scrutiny ones. Risk Meridian keeps a per-obligation checklist with status, notes, and evidence across § 219.11(c)/(e)/(f)/(g)/(h)/(i): human review of inputs and outputs, pause/restrict/disable capability, accuracy monitoring, a redress mechanism with a point of contact, public-facing disclosure, PII minimization and training, and security testing — each labeled Mandatory or Advisory to match the rule's modal verbs.
TRAIGA Reasonable-Care Record
TRAIGA presumes a person used reasonable care (§ 552.105(c)). Risk Meridian keeps the current, attributable record — classifications, assessments, disclosures, and oversight — that preserves that presumption, answers a civil investigative demand if the Attorney General asks (§ 552.103), and evidences the NIST AI RMF affirmative defense (§ 552.105(e)).
Hospital districts
Both government and healthcare — in one program
Hospital districts sit at the intersection of the public-sector and healthcare AI stacks. That overlap is a burden when it lives in spreadsheets and a strength when it lives in one documented, defensible record. Risk Meridian lets your district govern AI use across every applicable framework without duplicating work.
- A hospital district can be a governmental entity and a healthcare provider at the same time.
- Its AI use may span SB 1964, HB 3512, TRAIGA, and SB 1188 — often simultaneously.
- Applicability of each statute should be confirmed with counsel, not assumed.
- Document once across all four frameworks from a single system of record.
SB 1964 and HB 3512 apply to governmental entities; a hospital district may qualify. SB 1964 lets a district satisfy the standardized-notice disclosure with a generalized statement in its patient consent forms that AI may be used in treatment (§ 2054.711(c)). Whether SB 1188 and the others apply to your district should be confirmed with counsel.
AI governance for Texas government — FAQs
Common questions from agency IT leaders, procurement officers, hospital-district administrators, and public-sector counsel.
- Does SB 1964 apply to my agency or local government?
- SB 1964 adds Subchapter S to Government Code chapter 2054 and applies to state agencies and local governments that procure, develop, deploy, or use AI — it is not a mandate on private companies. Once in scope, an entity shall adopt the DIR AI Code of Ethics (§ 2054.702) and DIR minimum standards for Heightened Scrutiny AI systems (§ 2054.703), inventory its AI and HSAI with a purpose and risk-mitigation evaluation (§§ 2054.068, 2054.0965), conduct confidential HSAI impact assessments (§ 2054.708), and post the DIR standardized notice (§ 2054.711). Whether a specific special-purpose unit is a covered entity, and the exact content of the DIR-set code of ethics, standards, and forms, should be confirmed with counsel and against current DIR guidance.
- What does HB 3512 require, and who has to be trained?
- HB 3512 requires annual AI training for public-sector personnel. A state agency must identify each employee who uses a computer for at least 25% of their duties; those employees and every elected or appointed officer must complete a DIR-certified AI training program at least once a year (§§ 2054.5191(a), 2054.5193). Local governments carry the same duty for employees and officials with computer-system access (§ 2054.5191(a-1)). Completion is verified and reported to DIR on its form and is subject to periodic audits, and each state agency certifies compliance in its strategic plan (§ 2056.002(b)(12)). DIR certifies at least five AI-training programs; Risk Meridian records completion and expiry for the personnel you identify and produces that reporting.
- How does TRAIGA affect government agencies specifically?
- TRAIGA is an intent-based prohibition statute that comes into force January 1, 2026. Its general prohibitions apply broadly, and government agencies additionally have AI-use disclosure duties. TRAIGA is enforced exclusively by the Texas Attorney General, includes a 60-day cure period, and has no private right of action. Because NIST AI RMF substantial compliance is an explicit affirmative defense, keeping a documented governance record is how your agency evidences good faith.
- Our hospital district provides healthcare — which laws apply?
- A hospital district is a healthcare provider, so TRAIGA's patient AI-use disclosure duty (§ 552.051(f)) and SB 1188 apply. One nuance worth knowing: TRAIGA expressly excludes hospital districts from its “governmental entity” definition (Bus. & Com. Code § 552.001(3)), so TRAIGA's governmental-only prohibitions do not bind districts under that chapter — but SB 1964 and HB 3512 may reach them on their own terms. SB 1964 also gives districts a concrete break: an academic medical center, state-owned or public hospital, or hospital district may satisfy the standardized-notice disclosure by including a generalized statement in its patient consent forms that an AI system may be used in the course of treatment (§ 2054.711(c)). Two open items to track: DIR declined to confirm that the consent-form method also satisfies the separate disclosure duty in its Chapter 219 ethics rule (§ 219.11(g)(2)(C)), and Chapter 219 reaches local governments only in limited scope — whether it covers a given district turns on the rule's definitions (§ 219.1). Note also that public hospitals and districts carry heavier AI duties under SB 1964 than private competitors face under TRAIGA — commenters asked DIR to reconcile that asymmetry, and DIR said it lacks the statutory authority. Whether each statute reaches your district turns on the statutory definitions, so confirm applicability with counsel; Risk Meridian lets you document across all of them from a single system of record.
- What can I hand to procurement, my board, or an auditor?
- Risk Meridian produces PDF artifacts from your live records — Heightened Scrutiny determinations with written rationales (§ 2054.003), versioned § 219.22 risk assessments and § 219.23(b) impact assessments (mandatory for state agencies and universities under § 2054.708; advisory for local governments under 1 TAC § 219.23(e)(2)), the DIR Submission Pack for the local-government on-request review (§ 2054.0965(c)), the IRDR AI Answer Set for state agencies and universities (§ 2054.0965(b)(6)–(7)), the Classification Crosswalk reconciling TRAIGA-lineage and SB 1964/TAC 219 vocabularies, vendor-clause status, and HB 3512 training reports including the § 2054.5191(e) completion export and § 2056.002(b)(12) certification record. Every action is captured in a tamper-evident audit log with role-based access, so the evidence you present is attributable and defensible for procurement, board, and audit review.
- We are a public university — does this stack apply to us, and how does TRAIGA fit?
- Institutions of higher education are treated as state agencies under Government Code chapter 2054, so a public university carries the full state-agency tier of SB 1964 — the AI and HSAI inventory (§ 2054.068), the information-resources deployment review with per-system purpose, risk-mitigation, and strategic-plan-support evaluation plus the compliance confirmation (§ 2054.0965(b)(6)–(7)), confidential HSAI impact assessments (§ 2054.708), and the disclosure duties (§§ 2054.707, 2054.711) — along with the 1 TAC Chapter 219 workflow and HB 3512 training. At the same time, TRAIGA expressly excludes institutions of higher education from its “governmental entity” definition (Bus. & Com. Code § 552.001(3)), so TRAIGA’s governmental-entity-only rules do not bind them under that chapter. Risk Meridian models exactly that split for Institution of Higher Education organizations. Confirm your institution’s status with counsel.
- Is Risk Meridian TX-RAMP certified?
- Not yet. Risk Meridian is preparing its TX-RAMP Level 2 package. Texas agencies may sponsor provisional certification and can contract during the provisional period. Two practical notes: first, get audit-ready now — agency-sponsored provisional certification lets you contract while full certification completes. Second, the compliance work itself is procurement-independent: your inventory, assessments, and § 2054.0965 filings are your own obligations, and your team can prepare them regardless of any vendor’s certification status.
- Does Risk Meridian guarantee compliance with these statutes?
- No. Risk Meridian does not guarantee compliance, and no vendor can. What the platform does is help you document your governance program and build a defensible record across SB 1964, HB 3512, TRAIGA, and — for hospital districts — SB 1188. Specific obligations and applicability should always be confirmed against the statutes and with your counsel.
Get your public-sector AI program procurement-ready
Start now. Classify your first governmental AI system, capture the rationale, and begin building the DIR-ready, board-ready, audit-ready record your agency needs — audit-ready in under an hour.
No credit card required.
SB 1964, HB 3512, and TRAIGA support in one platform
Governmental-AI classifier with written rationales
Encrypted · RBAC · Tamper-evident audit log