Texas TRAIGA in Practice — 6 Months Post-Effective

The Texas Responsible Artificial Intelligence Governance Act (TRAIGA / HB 149) has been effective since September 1, 2025. Six-plus months in, this post summarizes how compliance teams are operationalizing the law and what early enforcement signals look like.

What TRAIGA actually requires

The core obligations under Tex. Bus. & Com. Code Ch. 49:

Disclosure when consumers interact with AI — clear and conspicuous disclosure when a reasonable consumer might believe they are interacting with a human (the chatbot/voice-bot scope)
Prohibition on AI developed for unlawful intentional discrimination under Texas or federal law
Restrictions on certain government AI uses — social scoring, biometric ID without consent, certain elections-related uses
30-day cure period under § 49.451(c) for many curable violations
$10,000 maximum civil penalty per violation under § 49.452(d)

Note: TRAIGA's scope is narrower than the Colorado AI Act. It does not impose comprehensive impact-assessment requirements like Colorado's § 6-1-1703(3).

What enforcement has looked like

*The following reflects general industry observations and is not legal advice or enforcement-strategy guidance.*

Disclosure focus: early enforcement attention has reportedly focused on chatbot-disclosure failures — entities deploying AI in customer-facing channels without clear disclosure that the customer is interacting with AI.
Cure-period leverage: the 30-day cure provision has been used effectively by deployers to bring disclosure surfaces into compliant shape after AG notice without penalty assessment.
No public discrimination-focused TRAIGA actions yet: as of April 2026 there have been no widely-reported AG enforcement actions under TRAIGA's discrimination-prohibition provision.

What compliance teams have learned

1. Audit your chatbot disclosures first

If your customer-facing channels include any AI, the disclosure surface is the immediate compliance priority. Verify:

The disclosure appears at the start of the interaction
It is clear ("You are chatting with an AI assistant" or equivalent)
It is conspicuous (not buried in a footer or modal that closes)
It survives across handoffs (if the chat is escalated to a human, the consumer should know the change)

2. Inventory automated decisions for discrimination risk

TRAIGA's prohibition is narrower than Colorado's high-risk regime — it targets intentional discrimination. But "intentional" includes patterns where a reasonable developer should have known the system would produce discriminatory outputs. Inventory your systems and run bias testing before regulatory inquiry.

3. Build the cure response

When TRAIGA notice arrives, the 30-day cure clock starts. Have a process: who triages, what evidence to gather, who can authorize remediation, how to document the cure.

How TRAIGA fits with Colorado, NIST, ISO

For multi-state deployers:

TRAIGA's narrower scope means a Colorado-AI-Act-grade compliance program will substantially exceed TRAIGA requirements
NIST AI RMF MEASURE function bias-testing supports TRAIGA's intentional-discrimination defensibility
ISO/IEC 42001 Annex A.4 + A.7 policies and data-management controls cover TRAIGA's governance scope

What to watch

The first widely-reported TRAIGA enforcement action against discrimination patterns will set the practical standard for the prohibition's reach
Possible expansion via amendment in subsequent Texas legislative sessions
Cross-pollination as other state AGs adopt TRAIGA-style narrower scopes vs Colorado-style comprehensive