The Social Services Appropriations Subcommittee voted to reduce the one-time appropriation for the Utah Health Artificial Intelligence Vault after members voiced concerns about re-identification and the absence of statewide AI guardrails.
Representative (speaker 32) argued that the state's population-scale data holdings (including driver license, PKU infant samples and all-payer claims) increase the risk that AI tools could re-identify individuals and amplify linkage risks. The member asked for a pause to request a comprehensive data-governance plan and administrative controls before committing the full appropriation.
Staff explained the request's history: the original ask was $18.6 million, the chairs reduced it to $12.5 million in the packet, and the committee motion further reduced it to $6.25 million while adding legislative intent language that a governance study be provided.
The motion to reduce passed; members said they wanted greater clarity about policy, disposition, opt-out requirements and the mitigation measures the department would use before additional appropriation.