The Texas House has given final approval to one of the most closely watched artificial intelligence bills of the year, potentially providing a roadmap for other red states.
The vote Friday came despite a looming threat that Congress could preempt enforcement of state-level AI regulations for the next decade as part of the federal budget reconciliation bill.
“This landmark legislation establishes the nation’s most robust, outcomes-based AI regulatory framework, balancing innovation with accountability to protect Texans and promote responsible AI leadership,” Rep. Giovanni Capriglione (R), the bill’s prime sponsor, tweeted after the vote.
If Gov. Greg Abbott (R) signs the bipartisan bill, Texas will follow states including Kentucky, Maryland and Montana that have restricted government use of AI, as well as states including California, Colorado and Utah that have placed obligations on AI developers and deployers in the absence of federal regulations.
The Texas measure, known as the Texas Responsible AI Governance Act, took a winding path to final passage. When it was first introduced in December it borrowed heavily from the Colorado AI Act, a 2024 law focused on preventing algorithmic discrimination by AI systems.
But in March, Capriglione radically overhauled the bill in response to criticism from business and free market groups. The new bill shifted the focus to government use of AI while still addressing a specific set of concerns related to private sector deployment.
The revamped bill’s stated purpose is to facilitate the responsible use of AI, protect individuals from AI’s risks, ensure transparency, and require that government agencies disclose to consumers when they are interacting with an AI system. It no longer references algorithmic discrimination.
Instead, the bill bans government agencies from using AI for “social scoring” if it would have a detrimental effect on an individual or group.
Social scoring is the practice of categorizing people based on their characteristics or social behavior. An example, according to Human Rights Watch, would be the use of an AI scoring system to determine if someone is a fraud risk for public benefits.
The bill also prohibits government agencies from using AI without consent to identify individuals using biometric data or from images gathered from the internet.
On the private sector side, the bill bars the development or deployment of an AI system that “intentionally aims to incite or encourage” someone to engage in self-harm, hurt someone else or commit a crime.
The bill would also:
- Prohibit AI systems built with the intent to discriminate against protected groups.
- Ban AI systems that are designed with the “sole intent” to make sexually explicit content or chatbots that engage in underage sexualized conversations.
- Preempt local AI ordinances and regulations.
To encourage AI development, the bill includes a regulatory “sandbox program” to allow developers up to 36 months to test their systems with protection from legal liability or regulatory enforcement. It also creates a Texas Artificial Intelligence Council to “ensure that artificial intelligence systems in [Texas] are ethical and developed in the public’s best interest,” among other responsibilities.
Texas’s attorney general would enforce the law. Companies would have 60 days to “cure” violations before facing penalties of up to $200,000 per violation. The law would take effect in January.
The Transparency Coalition, a Seattle-based nonprofit that advocates for AI safeguards, praised the legislation and called on Abbott to sign it.
“This bill shows that AI regulation is not a partisan issue and can benefit all citizens,” Jai Jaisimha, the coalition’s co-founder said in a statement.
“This milestone is an important first step in balancing the needs of society with deriving the full benefits of safer and more responsible AI,” he added.
Adam Thierer, senior research fellow at the free market think tank R Street Institute, said the bill had improved from its earlier version but “still represents excessive regulation of AI commerce and speech, issues better addressed at the federal level anyway.”
Thierer predicted that, if enacted, the law will be challenged in court.
Passage of the Texas bill comes as state-level momentum to pass AI regulations face headwinds amid growing concern about international competition and pushback from venture capitalists, Congress and the Trump administration.
A multistate effort to enact laws similar to Colorado’s have so far come up short, although bills are still pending in states including California, Connecticut and New York.
The Texas bill represented a departure from that approach, which Capriglione said in his statement will ensure “a future where AI serves the public good.”