Utah Tests AI Regulation Through Sandbox for Emerging Tech

Utah's experimental approach to AI oversight

Utah has established a unique framework for understanding and regulating artificial intelligence by allowing companies to test emerging AI systems under state supervision before formal rules exist. The state's Office of Artificial Intelligence Policy operates both a learning lab and a regulatory sandbox—a controlled environment where businesses can deploy untested products while regulators observe and learn.

Zach Boyd, who leads the office, explained that agreements with participating companies permit them to "go forward and potentially beyond what current regulations specify," giving lawmakers concrete data before establishing permanent rules. The approach reflects a deliberate strategy: foster innovation, protect the public, and proactively study a rapidly evolving technology, according to Margaret Busse, who heads the Utah Department of Commerce.

From sandbox to statute

The experimental framework has already influenced state law. One learning lab project led directly to HB452 during the 2025 legislative session, which created new requirements and consumer protections for AI in mental health care. The office also informed 2026 legislation addressing AI-generated deepfakes.

Currently, the office is examining AI agents—autonomous software systems that use artificial intelligence to plan and execute multistep actions with minimal human oversight. Boyd cited potential medical applications, including AI agents that could gradually adjust prescription dosages. The agency has received dozens of applications from leading companies seeking to deploy clinical AI systems.

"We're trying to figure out how to prudently clear some of those," Boyd said, "so that we can still oversee them and learn from them."

Why it matters

Utah's model offers a middle path between premature regulation that could stifle innovation and a hands-off approach that leaves consumers unprotected. By observing AI systems in controlled deployment, the state can craft evidence-based policies rather than reacting to problems after they emerge. The healthcare focus is particularly significant, as medical AI applications carry both substantial promise and serious risk. Boyd expects to deliver formal policy recommendations in October and November.

Lawmakers established the office in 2024 under the Artificial Intelligence Policy Act, originally set to expire in July 2027. The program will now continue under periodic legislative review to assess its effectiveness.

These details were first reported by The Salt Lake Tribune.