Texas TRAIGA + ISO 42001
In force since January 1, 2026, with a defense built on recognized frameworks
Bottom line
The Texas Responsible AI Governance Act (TRAIGA) has applied since January 1, 2026 to anyone doing business in Texas or developing or deploying AI there. It bans a short list of intentional harms rather than regulating risk broadly. Two provisions make ISO 42001 practical: a defense for violations found through internal review while substantially complying with the NIST GenAI Profile or another recognized AI risk management framework, and Attorney General demands for documentation that maps almost line for line onto 42001's documentation controls.[1][2]
What TRAIGA requires
For all developers and deployers, TRAIGA prohibits AI developed or deployed with intent to incite self-harm, harm to others or crime; AI developed with intent to unlawfully discriminate against a protected class (disparate impact alone is not enough); and AI made to produce child sexual abuse material or unlawful deepfakes. Licensed health care providers must disclose AI use in services or treatment.[1][2]
Government agencies carry extra duties: disclose AI to consumers before or at the time of interaction, no social scoring, and limits on biometric identification.[1]
The law also creates a 36-month regulatory sandbox and the Texas Artificial Intelligence Council.[1]
- Statute
- HB 149, Tex. Bus. & Com. Code ch. 551–554
- Effective
- January 1, 2026
- Enforcement
- Attorney General only; no private right of action
- Cure period
- 60 days after written notice
- Civil penalties
- $10k–$12k curable; $80k–$200k uncurable; $2k–$40k per day continuing
- Framework reference
- NIST GenAI Profile or another recognized framework
Key dates
- June 22, 2025HB 149 signed[2]
- Jan 1, 2026TRAIGA takes effect[1]
- Sep 1, 2026Statutory deadline for the AG's online complaint mechanism (§552.102)[1][3]
- 2027Next regular legislative session, the earliest window for amendments
Where they overlap
Each row is a TRAIGA provision and where an ISO 42001 AI management system provides the process or evidence. Section numbers refer to the Business & Commerce Code as added by HB 149.[1]
| Texas TRAIGA | ISO 42001 | Coverage |
|---|---|---|
| §552.105(e) Defense: internal review + substantial compliance with a recognized framework | Whole AIMS, 9.1, 9.2, 10.2 If ISO 42001 counts as a recognized framework (not yet confirmed), a working AIMS supports substantial compliance, and its monitoring and internal audit supply the internal review. | Partial |
| §552.103 AG demand: purpose, intended use, deployment context, benefits | A.6.2.2, A.9.4, 6.1.4 System requirements, intended use and impact assessment. | Strong |
| §552.103 AG demand: training data and input categories | A.7.2–A.7.6 Data acquisition, quality, provenance and preparation records. | Strong |
| §552.103 AG demand: outputs, metrics, known limitations | A.6.2.4, A.6.2.7, A.8.2 Verification and validation results, technical documentation and user information. | Strong |
| §552.103 AG demand: post-deployment monitoring and safeguards | A.6.2.6, A.9.2, 9.1 Operation and monitoring, responsible-use processes, performance evaluation. | Strong |
| §552.104 Notice and 60-day cure with written statement | 10.2, 7.5 Corrective action with documented evidence is exactly what a cure statement needs. | Strong |
| §552.052 No AI intended to incite self-harm, harm or crime | A.9.4, A.6.1.2, 6.1.4 Intended use and misuse analysis document design intent. Intent is ultimately a legal question. | Partial |
| §552.056 No AI developed with intent to unlawfully discriminate | A.5.4, A.7.4 Impact on individuals and data quality controls evidence non-discriminatory design. | Partial |
| §552.057 No AI made for CSAM, unlawful deepfakes or minor-impersonating sexual chatbots | A.6.2.2, A.9.2 Requirements and responsible-use processes; content safeguards are technical controls you implement. | Partial |
| §552.051 AI disclosure (government agencies; health care providers) | A.8.2, A.8.5 Information for users and interested parties; the clear-and-conspicuous standard is statutory. | Partial |
Coverage reflects how directly ISO 42001 produces the evidence or process the requirement asks for. It is an editorial assessment, not a legal opinion or a presumption of conformity.
What ISO 42001 won't cover
ISO 42001 makes the defense and the AG's questions much easier to answer. Some things it can't settle.
ISO 42001 is not named
The defense names the NIST GenAI Profile and any other nationally or internationally recognized framework. ISO 42001 fits that description, but no AG guidance confirms it.[1]
The defense is narrow
It applies to violations you discover through internal review, feedback or testing, not as blanket immunity for compliance.[1]
Content safeguards
Preventing CSAM, deepfakes or harmful outputs needs technical guardrails and red-teaming that 42001 requires you to consider but doesn't specify.
Government-only duties
Social scoring and biometric limits apply to agencies. Vendors selling to Texas agencies should support them contractually through A.10.
Using ISO 42001 to get there
- 1
Scope Texas exposure
Flag AI systems offered to Texas residents, used in health care services, or sold to Texas government agencies.
- 2
Document intent and misuse
Record intended use (A.9.4) and foreseeable misuse in the impact assessment so design intent is evidenced if challenged.
- 3
Prepare an AG response pack
Assemble the §552.103 items from AIMS records: purpose, data, inputs, outputs, metrics, limitations, monitoring and safeguards, plus any other documentation the AG requests.
- 4
Make internal review continuous
Monitoring, red-teaming, feedback channels and internal audit are how violations get discovered on your terms, which is what the defense rewards.
- 5
Run corrective action to a cure standard
Use 10.2 to fix, document and evidence each finding so a written cure statement takes days, not the full 60.
- 6
Consider certification
Third-party certification is independent evidence of a working AI risk management system, useful if you rely on the defense.
Frequently asked questions
Does ISO 42001 count as a recognized framework under TRAIGA?
The statute names the NIST AI RMF Generative AI Profile and "another nationally or internationally recognized risk management framework for artificial intelligence systems." ISO/IEC 42001 is an international standard for AI management systems and fits that description, but it is not named and there is no AG guidance yet. Treat it as a strong argument, not a guarantee.
Does TRAIGA apply to private companies?
Yes. The prohibitions on intentional harms, intentional unlawful discrimination and CSAM or deepfake systems apply to any developer or deployer doing business in Texas. Disclosure, social scoring and biometric rules apply mainly to government agencies, and health care providers must disclose AI use.
What can the Texas Attorney General ask for?
After a complaint, the AG may issue a civil investigative demand for a description of the system's purpose and intended use, training data, inputs, outputs, performance metrics, known limitations, and post-deployment monitoring and safeguards.
What are the penalties?
After a 60-day cure period, civil penalties range from $10,000 to $12,000 per curable violation, $80,000 to $200,000 per uncurable violation, and $2,000 to $40,000 per day for continuing violations. There is no private right of action.
Is TRAIGA being enforced?
The AG's complaint portal is live. As of this review there are no public TRAIGA enforcement actions, though investigative demands are often not announced.
Related frameworks
This page is general information, not legal advice. Laws change; confirm obligations with counsel.