Texas Enacts TRAIGA: A New Blueprint for Responsible AI Governance

According to an article by DLA Piper, the Texas Responsible Artificial Intelligence Governance Act (TRAIGA), signed into law on June 22, 2025, and effective January 1, 2026, makes Texas the third state with comprehensive AI regulations. Any organization that markets, sells, or develops AI interacting with the state is covered. Lawmakers promise a pro-innovation approach, striking a balance between guardrails and protection against discrimination and manipulation, while providing procedural relief and an experimental sandbox.

Unlike the tiered EU model, TRAIGA assigns role‑based duties to “developers” and “deployers” and simply prohibits certain uses. Government agencies must clearly disclose their interactions with AI, while all actors must avoid systems designed to encourage self-harm, violence, or crime. Social scoring and biometric identification that erode constitutional rights are prohibited. Liability for discrimination requires proof of intent, not disparate statistical impact.

Exclusive enforcement rests with the Attorney General, who must offer a sixty-day cure period before initiating legal action; uncurable breaches can result in fines of up to USD 200,000. Multiple safe harbors protect firms that uncover flaws through red-team testing or adhere to frameworks such as NIST’s AI Risk-Management Profile. 

A 36-month sandbox, overseen by the Department of Information Resources, allows approved projects to test novel systems without requiring specific licenses, while statewide preemption avoids the creation of municipal patchworks.

For compliance and risk leaders, the eighteen‑month runway for the Texas Responsible Artificial Intelligence Governance Act is short. The article suggests compliance leaders should begin by inventorying AI use cases involving Texan data, users, or operations; document the shutdown of prohibited functions; embed adversarial testing, impact assessments, and traceable decision records; and align enterprise policies with NIST, ISO 42001, Colorado’s AI Act, and the forthcoming EU regime. 

Additionally, they should closely monitor a congressional proposal that would tie broadband funding to a decade-long moratorium on state AI laws, a development that could halt TRAIGA enforcement. Finally, consider applying to the sandbox to pilot compliant innovations.