What Texas's AI Regulatory Sandbox Offers Builders
By Matthew Bertram ·
Last reviewed June 25, 2026
TRAIGA does something most state AI laws don’t: it builds in a sanctioned testing lane. The law, effective January 1, 2026, authorizes the Texas Department of Information Resources to run an AI regulatory sandbox — a formal program that lets builders develop and test AI systems without first clearing separate state licensing or regulatory-authorization hurdles.
What the sandbox actually provides
Participants get two concrete protections during the testing period. First, the Attorney General cannot file enforcement charges for violations of state laws that are waived for sandbox purposes. Second, state agencies cannot impose fines or suspend licenses for those same waived requirements while testing is active. The period runs up to 36 months, and DIR can extend it for good cause.
Oversight sits with the Texas Artificial Intelligence Council, a seven-member body appointed by the governor, lieutenant governor, and speaker of the House, and administratively attached to DIR. The Council — or any applicable agency — can recommend removing a participant whose AI poses undue risk to public safety, violates federal law, or violates non-waived state law. DIR keeps trade secrets and other sensitive submissions confidential.
What applicants must show
There is no lightweight path in. Applicants submit four things: a detailed description of the AI system and its intended use; a benefit assessment covering consumer impact, privacy, and public safety; a plan for mitigating adverse consequences during testing; and proof of compliance with applicable federal AI laws and regulations.
Once accepted, participants file quarterly reports on performance metrics, risk-mitigation updates, and feedback from users and stakeholders. DIR, in turn, reports to the legislature each year with sandbox outcomes and policy recommendations.
The caveat that matters
The sandbox does not suspend TRAIGA’s core prohibitions. A company inside it still cannot deploy AI designed to manipulate users into self-harm or criminal activity, infringe constitutional rights, engage in intentional unlawful discrimination, or generate child sexual content or deepfakes. Those apply no matter your testing status.
As of June 2026, the sandbox is authorized by law and in effect. Whether DIR has published application forms or begun accepting participants hasn’t been confirmed — builders who want to apply should check directly with DIR at dir.texas.gov.
Frequently asked questions
What can trigger removal from the TRAIGA sandbox once a company is accepted?
The Texas Artificial Intelligence Council, or any applicable state agency, can recommend removing a participant if the AI system poses undue risk to public safety, violates federal law, or violates a state law that was not waived as part of participation. Removal is not automatic — it takes a recommendation from the Council or the relevant agency.
Does sandbox participation excuse a company from TRAIGA's core prohibitions?
No. The sandbox waives separate licensing and regulatory-authorization requirements, but TRAIGA's four core prohibitions stay in force regardless of participation. A sandboxed company still cannot deploy AI designed to manipulate users into self-harm or criminal activity, infringe constitutional rights, engage in intentional unlawful discrimination, or generate child sexual content or non-consensual deepfakes.
Who sits on the Texas Artificial Intelligence Council that oversees the sandbox?
The Council has seven members appointed across three offices — the governor, the lieutenant governor, and the speaker of the House. It is administratively attached to DIR and submits annual reports to the legislature with sandbox outcomes and policy recommendations.
Analysis and commentary, not legal advice.