Get Full Government Meeting Transcripts, Videos, & Alerts Forever!
Get email alerts on the Ai Governance topic
No spam. Unsubscribe anytime.
Wyoming committee debates human-review requirement for government use of AI, defers bill
Summary
A proposed bill that would bar consequential government decisions made by automated systems without meaningful human review prompted a long committee debate over scope, liability and operational exceptions; the panel deferred action for further drafting.
The select committee reviewed a short bill draft that would prohibit governmental entities from making "consequential decisions" via automated decision-making systems without "meaningful human review." The draft, presented by staff, included definitions of automated decision-making systems, consequential decisions, and a staff-proposed meaningful-human-review standard.
Staff advised the committee that other states vary in approach: some (California, Illinois) have adverse-action-focused rules, while others emphasize developer obligations or impact assessments. Committee members debated whether the Wyoming draft should cover any consequential decision (positive or negative) or be limited to adverse outcomes such as denials of benefits or licensing.
Proponents said the human-in-the-loop requirement is a best practice to prevent automated adverse outcomes that leave affected people without timely redress. "Make sure that if it's doing something of consequence, that a human actually had legitimately their eyes on it," the chair said, arguing it reduces avoidable harms.
Opponents cautioned against overreach and said existing legal remedies and grievance processes apply regardless of the automation layer. Senator Nethercott argued the bill risks hamstringing government operations and suggested focusing on accountability and remedies rather than defining AI broadly in statute.
The committee also heard public comment from Alan Thompson (Wyoming Association of Sheriffs and Chiefs of Police), who asked for exceptions to allow automated emergency interventions (for safety or fraud prevention) pending rapid human review.
Outcome: After extended discussion the committee declined to take action at the meeting and asked staff to refine language, consider carve-outs for emergency operational systems and examine whether to constrain the statute to adverse actions or to an automated-decision definition rather than an AI definition.
Why it matters: The debate highlights tension between preventing harmful automated decisions that affect benefits and rights and avoiding statutory language that could impede government efficiency or create unintended boundaries across agencies.
