Texas Launches an AI Sandbox—and It Could Be a Game Changer
Texas has officially joined the wave of states trying to wrangle artificial intelligence.
On June 22, 2025, the Texas Responsible Artificial Intelligence Governance Act (TRAIGA) was signed into law and takes effect January 1, 2026. It’s a big step toward regulating how AI is developed and used in Texas, especially by government agencies. But one of the most interesting features of the law isn’t just the restrictions—it’s the opportunity it offers through what’s called a regulatory “sandbox”.
Who TRAIGA Covers and What It Regulates
TRAIGA sets guardrails around the use of AI in Texas. It applies to developers, deployers, and even government entities that use AI in any way that affects Texas residents. However, employment or commercial uses are not subject to TRAIGA.
The law defines AI broadly. If your system takes inputs and produces content, decisions, or recommendations that can influence anything in the real or digital world, you might fall under its scope. There are also consumer disclosure requirements. If your AI interacts with people in Texas, the law says you have to clearly tell them that they’re dealing with AI even if it seems obvious.
On top of that, TRAIGA bans certain uses, including AI that discriminates, tracks people with biometric data without consent, or encourages self-harm or violence. Only the Texas Attorney General can enforce the law. No private lawsuits or class actions are allowed, but civil penalties for violations can reach six figures.
The Sandbox Opportunity
The part that’s turning heads, though, is the sandbox. TRAIGA creates an AI regulatory sandbox that gives companies a chance to test AI systems in a real-world environment for up to 36 months, without needing to fully comply with all Texas laws from day one. This is especially valuable for startups and innovators working on novel AI tools.
To participate, companies have to apply and explain how their AI works, what risks it poses, and how they plan to manage those risks. Once in the program, they’ll submit quarterly reports with performance data and feedback from users. The Attorney General and state agencies won’t bring enforcement actions while a company is in the sandbox, as long as it follows TRAIGA’s core restrictions.
No Federal Preemption
Earlier this year, Congress floated a proposal to block state-level AI laws for 10 years as part of the Big Beautiful Bill, but the moratorium failed to pass. That means there’s no federal override. TRAIGA is the law in Texas, and companies that use or develop AI here will need to follow it.