AI Policy Wiki
Dashboard

Texas Responsible AI Governance Act (TRAIGA / HB 149) — Source Summary

high confidence · updated 2026-06-06

Summary of Texas HB 149 (TRAIGA), signed June 22, 2025 and effective Jan 1, 2026 — the first comprehensive AI law enacted in a Republican-led state. Establishes prohibited uses, a regulatory sandbox, the Texas AI Advisory Council, and AG-exclusive enforcement with a 60-day cure period.

The Texas Responsible AI Governance Act (TRAIGA), enacted as HB 149, is described in the source as the first comprehensive AI governance statute enacted in a Republican-led US state. It was signed by Governor Greg Abbott on June 22, 2025, took effect January 1, 2026, and was primarily sponsored by Rep. Giovanni Capriglione. Enforcement rests exclusively with the Texas Attorney General, with no private right of action. The statute establishes a set of intent-based prohibited uses, a regulatory sandbox, and a non-binding advisory council.

Summary

The source states that TRAIGA was originally modeled on the Colorado AI Act (SB 24-205) but was scaled back during the legislative process. It dropped broad developer and deployer duties of care over "high-risk" systems in favor of a narrower set of intent-based prohibitions, a regulatory sandbox, and a non-binding advisory council. According to the source, framing the prohibitions around intent rather than outcomes narrows the statute's reach compared with Colorado's duty-of-care model.

Scope

The Act applies to persons who promote, advertise, or conduct business in Texas; produce a product or service used by Texas residents; or develop or deploy an AI system in Texas (§551.002).

Prohibited practices (Subchapter B)

  • Behavioral manipulation (§552.052): AI systems that "intentionally aim to incite or encourage" self-harm, harm to others, or criminal activity.
  • Social scoring by government (§552.053): Bars Texas governmental entities from using AI to evaluate persons "based on social behavior" in ways that yield "detrimental or unfavorable treatment" or rights infringement.
  • Constitutional-rights infringement (§552.055): Prohibits developing or deploying AI "with the sole intent ... to infringe, restrict, or otherwise impair" US constitutional rights.
  • Unlawful discrimination (§552.056): Prohibits AI developed or deployed "with the intent to unlawfully discriminate against a protected class." The source notes that "disparate impact is not sufficient by itself" to prove intent.
  • Child sexual abuse material (§552.057): Prohibits systems designed to generate CSAM or to impersonate a child under 18 in sexualized text conversations.
  • Agency disclosure (§552.051): Governmental agencies must disclose AI use to individuals before or at the time of interaction.

The source describes all of these prohibitions as framed around intent rather than outcomes, which it characterizes as substantially narrowing the statute's reach relative to Colorado's duty-of-care model (Source: this source).

Regulatory sandbox (Chapter 553)

Participants may test AI systems for up to 36 months, extendable for "good cause." During testing, the Attorney General may not file charges for violations of waived laws. The Subchapter B prohibitions (manipulation, discrimination, CSAM) cannot be waived. The source identifies this as the first US state AI sandbox and likens it to the UK FCA-style model, describing it as a counterweight to the prohibitions that gives firms a formal path to test novel deployments under AG supervision.

Texas Artificial Intelligence Advisory Council (Chapter 554)

The council has seven members: three appointed by the Governor, two by the Lieutenant Governor, and two by the Speaker. Members must be Texas residents with expertise in AI, data privacy, ethics, public policy, risk management, governmental efficiency, or anticompetitive practices. The council issues reports and conducts training but may not adopt binding rules or guidance (§554.103(1)). The source frames the non-binding limit as a deliberate contrast to Colorado, where the AG was empowered to issue implementing rules.

Enforcement

  • Exclusive AG enforcement (§552.101): no private right of action.
  • 60-day cure period (§552.104): after written notice, the AG cannot sue for 60 days if the violator cures the violation and certifies the cure in writing with supporting documentation and policy changes.
  • Civil penalties (§552.105): $10K–$12K (curable), $80K–$200K (uncurable), $2K–$40K per day (continuing); injunctive relief and fees available.

The source describes the 60-day cure window and AG-only enforcement posture as borrowed from Colorado's model, paired with a lower duty-of-care surface, producing a regime it characterizes as narrower but better-resourced for remediation.

Comparison with Colorado AI Act

DimensionColorado SB 24-205Texas TRAIGA
Trigger"High-risk" AI making consequential decisionsIntent-based prohibited uses
Who is regulatedDevelopers + deployers in covered domainsAny developer/deployer intentionally causing prohibited harm
Discrimination standardReasonable care; disparate impact actionableIntent required; disparate impact insufficient
Duty of careYes (impact assessments, risk management, notice)No general duty of care
Rulemaking authorityYes (AG)No (advisory council non-binding)
SandboxNoneYes, up to 36 months
EnforcementAG, 60-day cure, no private actionAG, 60-day cure, no private action
Political framingEquity / anti-discriminationInnovation-friendly guardrails

Relationships

Provenance

No PII, credentials, or sensitive identifiers in the source.