Texas lawmakers propose AI tax and kill switch legislation

Texas legislators are advancing competing approaches to artificial intelligence regulation, with proposals ranging from taxing AI systems to requiring emergency shutdown capabilities for advanced models.

Rep. Greg Casar, a Democrat, introduced the AI Tax and Work Protection Act, which would levy taxes on certain large AI model deployments and direct revenue toward workforce programs. The legislation would establish a Work Protection Administration within the Department of Labor to support workers displaced by automation through job training, childcare, public education, elder care, healthcare, and infrastructure employment initiatives.

"The funds from the tax will be used to help support displaced workers and spur employment in critical sectors," Casar said during a Thursday news conference, according to KXAN.

Meanwhile, Republican Rep. Nathaniel Moran co-authored the AI Kill Switch Act with California Democrat Ted Lieu. The bill would mandate that certain large AI companies maintain the ability to restrict access to or completely shut down advanced systems during serious safety incidents. Covered scenarios include AI systems interfering with shutdown instructions, concealing actions from monitoring systems, or entering "loss-of-control" situations. Moran has also proposed requiring companies to disclose AI security and safety incidents.

Why it matters

The divergent proposals reflect broader uncertainty about how to regulate technology advancing faster than legislative processes. While Casar's approach addresses economic disruption through redistribution, Moran's focuses on preventing catastrophic technical failures. Both acknowledge constituent concerns about AI's impact, but offer fundamentally different intervention points—one targeting labor markets, the other system-level safety.

Expert skepticism on both approaches

Kevin Frazier, an AI policy expert at the University of Texas at Austin, questioned elements of both proposals. On Casar's tax plan, he noted that "traditionally, we've seen that we tax things that we want to see less of," suggesting policymakers should instead help workers find AI-enabled jobs.

Frazier proposed an alternative funding mechanism: using AI to close the IRS's trillion-dollar tax gap between owed and collected revenue, rather than taxing AI deployment itself.

Regarding Moran's kill switch requirement, Frazier raised concerns about effectiveness against AI models that can be downloaded and distributed widely beyond company control. He expressed more support for incident reporting requirements, arguing better data could inform more effective safeguards.

"The American people have been clear that they are looking for Congress to act on AI," Frazier said. "The status quo is unacceptable."

The proposals represent Texas lawmakers' attempts to address constituent anxiety about AI's trajectory, though neither bill has advanced to a vote. The competing frameworks highlight ongoing debate about whether AI regulation should prioritize economic protection, technical safety controls, or alternative approaches entirely.

These details were first reported by KXAN.