Blab AI
Wyoming Senators Clash Over Draft AI Bill That Aims to Keep Humans in the Loop
The Wyoming Legislature’s Select Committee on Blockchain, Financial Technology and Digital Innovation Technology is debating a draft artificial‑intelligence bill that would restrict state agencies from making consequential decisions without human oversight. The bill, drafted during the 2024 budget session, draws from similar proposals in New Jersey, Colorado and Illinois.
The core of the proposal is a definition of “consequential decision making” – decisions that affect a person’s rights, benefits or legal obligations. Under the bill, any automated system that could influence such outcomes would be prohibited from acting independently. Agencies would also have to keep a record of which employees use these systems.
Only two committee members spoke on the bill. Sen. Chris Rothfuss, a Democrat from Laramie, said the legislation is needed to prevent the kind of errors that have already occurred in other states. He cited a lawsuit against UnitedHealth in which the insurer allegedly used faulty AI to deny coverage to elderly patients, and reporting from KUOW that Washington state may soon deny Medicare claims through AI. “The bill is trying to make sure that there's a human in control of key government decisions,” Rothfuss said.
Rothfuss added that the AI industry’s best practice is to have a human review final steps that could be consequential or adverse. He argued that the bill would not regulate how agencies implement AI, only that a human must be the last step in the decision chain.
Sen. Tara Nethercott, a Republican from Cheyenne, opposed the bill. She said it was “in search of a problem” and that technology does not hide wrongdoing. “Regardless of the process used by any government entity, or other entity, regarding any decision it's made, the company or government entity doesn't get to not have liability due to the process it used to make the decision,” Nethercott said. She warned that the bill’s definition of artificial intelligence could have broad implications beyond government use.
Nethercott also suggested that a complete ban on government use of AI or a specific appeals process for technology‑related errors might be more effective. She said the bill fails to address the underlying fears about technology.
The committee’s debate highlighted a split over liability and oversight. Rothfuss stressed that the bill would prevent situations where a person was eligible yesterday but denied today because an AI made the decision without a human review. “It’s not about liability because these are individual benefits. So yes, if we brought a court in, it would get resolved,” he said.
The bill defines an artificial‑intelligence system as “any machine‑based system that … infers from the inputs the system receives how to generate outputs … that can influence physical or virtual environments.” The proposal would require state agencies to keep a log of employees who operate such systems.
After the exchange, the committee laid the bill back for now. The Select Committee on Blockchain, Financial Technology and Digital Innovation Technology will reconvene on September 28 at the University of Wyoming in Laramie to revisit the proposal.
At this stage, no legislative action has been taken. The bill remains a draft, and its future depends on the committee’s next meeting and the broader legislative agenda for the 68th Wyoming Legislature.
The debate reflects a growing national conversation about how to balance the benefits of AI with safeguards that protect citizens from erroneous automated decisions. Wyoming’s approach, if adopted, would be among the first state‑level attempts to codify human oversight in AI‑driven public services.
The bill’s status is currently uncertain, and it will be closely watched by stakeholders in the state’s healthcare, social services and technology sectors as the committee prepares for its next session.
The core of the proposal is a definition of “consequential decision making” – decisions that affect a person’s rights, benefits or legal obligations. Under the bill, any automated system that could influence such outcomes would be prohibited from acting independently. Agencies would also have to keep a record of which employees use these systems.
Only two committee members spoke on the bill. Sen. Chris Rothfuss, a Democrat from Laramie, said the legislation is needed to prevent the kind of errors that have already occurred in other states. He cited a lawsuit against UnitedHealth in which the insurer allegedly used faulty AI to deny coverage to elderly patients, and reporting from KUOW that Washington state may soon deny Medicare claims through AI. “The bill is trying to make sure that there's a human in control of key government decisions,” Rothfuss said.
Rothfuss added that the AI industry’s best practice is to have a human review final steps that could be consequential or adverse. He argued that the bill would not regulate how agencies implement AI, only that a human must be the last step in the decision chain.
Sen. Tara Nethercott, a Republican from Cheyenne, opposed the bill. She said it was “in search of a problem” and that technology does not hide wrongdoing. “Regardless of the process used by any government entity, or other entity, regarding any decision it's made, the company or government entity doesn't get to not have liability due to the process it used to make the decision,” Nethercott said. She warned that the bill’s definition of artificial intelligence could have broad implications beyond government use.
Nethercott also suggested that a complete ban on government use of AI or a specific appeals process for technology‑related errors might be more effective. She said the bill fails to address the underlying fears about technology.
The committee’s debate highlighted a split over liability and oversight. Rothfuss stressed that the bill would prevent situations where a person was eligible yesterday but denied today because an AI made the decision without a human review. “It’s not about liability because these are individual benefits. So yes, if we brought a court in, it would get resolved,” he said.
The bill defines an artificial‑intelligence system as “any machine‑based system that … infers from the inputs the system receives how to generate outputs … that can influence physical or virtual environments.” The proposal would require state agencies to keep a log of employees who operate such systems.
After the exchange, the committee laid the bill back for now. The Select Committee on Blockchain, Financial Technology and Digital Innovation Technology will reconvene on September 28 at the University of Wyoming in Laramie to revisit the proposal.
At this stage, no legislative action has been taken. The bill remains a draft, and its future depends on the committee’s next meeting and the broader legislative agenda for the 68th Wyoming Legislature.
The debate reflects a growing national conversation about how to balance the benefits of AI with safeguards that protect citizens from erroneous automated decisions. Wyoming’s approach, if adopted, would be among the first state‑level attempts to codify human oversight in AI‑driven public services.
The bill’s status is currently uncertain, and it will be closely watched by stakeholders in the state’s healthcare, social services and technology sectors as the committee prepares for its next session.
