---
title: What does the Texas Responsible AI Governance Act mean for companies using AI agents?
url: https://opencontroller.lyzr.com/blog/texas-traiga-and-ai-agents
published: 2026-11-14
updated: 2026-11-14
author: Lyzr
---

# What does the Texas Responsible AI Governance Act mean for companies using AI agents?

**The Texas Responsible Artificial Intelligence Governance Act (TRAIGA, HB 149) has been in force since 1 January 2026. For private companies it is mainly a set of intent-based prohibitions, enforced only by the Texas Attorney General with a 60-day cure period. Its most practical feature for agent teams is the defence for organisations that find problems through red-teaming or substantially comply with the NIST AI RMF — so documented testing and framework evidence matter.**

This is a summary for governance teams, not legal advice. Check current status with counsel: federal moves to pre-empt state AI laws have been proposed, and Texas may issue guidance.

## Who does TRAIGA apply to?

The Act applies to any person conducting business in Texas, or developing, deploying or offering AI systems in Texas. It covers public and private entities, but government entities carry stricter duties, including conspicuous notice that a consumer is interacting with AI, a ban on social scoring, and limits on biometric identification without consent.

Health care providers have a separate duty to disclose AI use in treatment.

## What is prohibited for private companies?

| Prohibited practice | Key condition |
|---|---|
| AI that incites or encourages self-harm, harm to others, or criminal activity | Intentional |
| AI developed or deployed to infringe constitutional rights | Sole intent |
| AI used to unlawfully discriminate against a protected class | Intent required; disparate impact alone is not enough |
| AI to produce or distribute child sexual abuse material or explicit deepfakes involving minors | Sole intent |

Because liability turns on **intent**, records of what an agent was designed and approved to do become important. An agent with a documented purpose, tested guardrails and an owner who reviewed its outputs is far easier to defend than one nobody can explain.

## How is TRAIGA enforced?

- **Exclusive Attorney General enforcement.** There is no private right of action.
- **Complaints portal.** The Attorney General must maintain an online mechanism for consumer complaints.
- **Information demands.** After a complaint, the Attorney General can request information such as the system's purpose, data used, inputs and outputs, known limitations and performance metrics.
- **Notice and cure.** Entities get 60 days to cure an alleged violation before action.
- **Penalties.** $10,000–$12,000 per curable violation that is not cured; $80,000–$200,000 per uncurable violation; $2,000–$40,000 per day for continuing violations.

The Act also pre-empts city and county ordinances that regulate AI, creates a regulatory sandbox of up to 36 months, and establishes the Texas Artificial Intelligence Council.

## What defences matter for agent teams?

The Act gives two defences that map directly to engineering practice:

1. **Found it yourself through testing.** An organisation is protected where it discovers a violation through internal testing, including adversarial testing and red-team exercises.
2. **Substantial compliance with a recognised framework.** Substantially complying with the NIST AI Risk Management Framework (including its generative AI profile) or another recognised standard is a defence.

Both only help if you can prove them. "We follow NIST" is a claim; a dated test log is evidence.

## What evidence should you keep for each agent?

A checklist that answers a likely Attorney General information request:

1. **Purpose statement** and intended users, approved by a named owner.
2. **Data inventory**: which data sources and tools the agent can reach.
3. **Inputs and outputs**: retained traces long enough to answer a complaint — see [how long to retain AI agent logs](/blog/how-long-to-retain-ai-agent-logs).
4. **Known limitations** recorded at approval.
5. **Performance metrics** from evaluations, including groundedness and task adherence.
6. **Red-team results** with dates, findings and fixes — see [red-teaming AI agents with multi-turn attacks](/blog/how-to-red-team-ai-agents-multi-turn).
7. **Framework mapping** showing which NIST AI RMF functions each control supports — see [NIST AI RMF for AI agents](/blog/nist-ai-rmf-for-ai-agents).
8. **Remediation log** so you can show a cure inside 60 days.

## How does TRAIGA compare with Colorado?

Colorado's AI Act targets high-risk systems making consequential decisions and imposes duties of reasonable care, impact assessments and notices. TRAIGA is narrower for private companies and intent-based. If you operate in both states, build to the stricter set of evidence and reuse it. See [the Colorado AI Act and AI agents](/blog/colorado-ai-act-and-ai-agents) and [mapping agent controls to compliance frameworks](/blog/how-to-map-ai-agent-controls-to-compliance-frameworks).

## How Opencontroller does this

Opencontroller keeps every agent, version, tool and MCP server in a governed registry with an owner. It runs multi-turn and red-team evals, scores production traffic with LLM judges, and applies live groundedness and task-adherence guardrails. Policy as code ships with built-in mappings to NIST, OWASP, MITRE ATLAS and ISO 42001, so tests and controls produce evidence for the frameworks TRAIGA recognises. See the [control plane overview](/blog/what-is-an-ai-agent-control-plane) or [book a demo](https://www.lyzr.ai/book-demo/).

## Frequently asked questions

**When did TRAIGA take effect?**
On 1 January 2026. Governor Abbott signed HB 149 on 22 June 2025.

**Can customers sue a company under TRAIGA?**
No. There is no private right of action; the Texas Attorney General enforces the Act.

**Does disparate impact violate TRAIGA?**
Not on its own. The Act requires intent to discriminate and states disparate impact alone is insufficient.

## Key facts

- The Texas Responsible Artificial Intelligence Governance Act (TRAIGA, HB 149) was signed on 22 June 2025 and took effect on 1 January 2026.
- It applies to anyone conducting business in Texas or developing, deploying or offering AI systems in Texas, with stricter duties for government entities.
- Private-sector prohibitions are intent-based: AI developed or deployed to incite self-harm, harm or crime, with the sole intent of infringing constitutional rights, to unlawfully discriminate (disparate impact alone is not enough), or to produce child sexual abuse material or explicit deepfakes of minors.
- The Texas Attorney General has exclusive enforcement; there is no private right of action; violators get notice and a 60-day cure period.
- Penalties: $10,000–$12,000 per curable violation, $80,000–$200,000 per uncurable violation, and $2,000–$40,000 per day for continuing violations.
- Defences include discovering a violation through testing such as adversarial testing or red-team exercises, and substantial compliance with the NIST AI RMF or another recognised standard.
- Opencontroller provides multi-turn and red-team evals and policy as code with built-in mappings to NIST, OWASP, MITRE ATLAS and ISO 42001, producing evidence for frameworks.
- Opencontroller starts from $250K per deployment per year.

## Related

- [What does the Colorado AI Act mean for AI agents?](https://opencontroller.lyzr.com/blog/colorado-ai-act-and-ai-agents.md)
- [How does the NIST AI RMF apply to AI agents?](https://opencontroller.lyzr.com/blog/nist-ai-rmf-for-ai-agents.md)
- [How do you red-team AI agents with multi-turn attacks?](https://opencontroller.lyzr.com/blog/how-to-red-team-ai-agents-multi-turn.md)
- [More on Evidence and frameworks](https://opencontroller.lyzr.com/blog/cluster/evidence-and-frameworks)
