Utah Governor Spencer Cox signed Executive Order 2026-08 on October 6, 2026. It sets how executive branch agencies adopt AI, under the “pro-human” principles of an initiative Cox launched in December 2025. The order took effect immediately and stays in force until it is modified or rescinded.
Training comes first. Each agency must ensure its applicable employees take AI literacy training prescribed by the Division of Technology Services and the Division of Human Resource Management. The two divisions must build a strategy that covers foundational AI concepts, practical use of the state’s approved tools, and the pro-human principles. Designated agency AI leads get specialized systems-analysis training. The technology division must also evaluate training platforms and content providers, giving priority to Utah-based ones.
The order sets guardrails for use. The Department of Government Operations helps agencies find where AI can be used responsibly, with human review for determinations that affect individual rights. Agencies must document how AI systems are built into processes that affect the public, including their inputs, outputs, and role in decisions. Agencies must avoid AI technologies that do not follow the pro-human principles.
Agencies must also look for work AI can do. Each must evaluate business functions that AI could improve in a cost-effective way and propose citizen-facing processes that could be automated responsibly. The Department of Government Operations reports progress to the governor by July 1, 2027, and every July 1 after that.
The order leaves two things open. It sets no deadline for the training. And it does not define which determinations count as affecting individual rights.
Training is the order’s first requirement for every agency. The first progress report is due July 1, 2027.
Why it matters
Utah’s order puts the workforce first. Making training a requirement for every agency treats AI as a skills problem as much as a technology problem. It also makes training a budget line. The order already directs the state to evaluate training vendors.
The human review clause is the floor, not the plan. It protects people only for the decisions it covers, and the order leaves that list to the agencies. The documentation requirement does more quiet work. An agency that records inputs, outputs, and the AI’s role for each public-facing process has most of the inventory an audit would need.
Questions for your agency
- Which employees count as “applicable,” and what does each role need to know?
- Does training cover the tools your staff actually use, or AI in general?
- Which of your decisions affect individual rights, and who is the human reviewing them?
- Could you document inputs, outputs, and the AI’s role for each public-facing process today?
- What will you be able to report a year from now?
Human review is only as strong as the list of decisions it covers. Utah’s order leaves that list to the agencies.
Sources
- Executive Order 2026-08, Establishing Utah’s Pro-Human Approach to Artificial Intelligence in State Government (Office of the Governor, Oct. 6, 2026)
- Utah governor’s “pro-human AI” order directs state government to use “responsible” artificial intelligence (Route Fifty)
- Utah governor signs executive order for pro-human AI (GovTech)
- Utah governor signs order for pro-human AI integration into state government (Deseret News, Oct. 6, 2026)