Kotek Orders Oregon to Develop Safety Standards for Advanced AI Used by State Agencies

SALEM, Ore. - Gov. Tina Kotek signed an executive order Wednesday directing Oregon to develop new safety standards for advanced artificial intelligence systems purchased or used by state agencies.

Executive Order 26-26 sets a policy for the state to seek out “frontier” AI models that have undergone independent safety review. It gives Oregon’s chief information officer 90 days to propose how the state would put that policy into practice.

The proposal must address questions the order leaves open, including what qualifies as a frontier AI model, what would count as an adequate review, and how the state would assess a reviewer’s independence and qualifications. It must also consider how safety reviews would fit into state purchasing and whether exceptions would be needed when an applicable review is unavailable.

Kotek also directed the chief information officer to evaluate whether a “kill switch” requirement would be viable. The order does not require one now. That idea is among the measures the state will study as it develops its proposal.

In announcing the order, Kotek said states need to address AI safety while lawmakers consider broader standards. “States, in the meantime, must act with more urgency and do everything possible to manage how AI is impacting our communities,” she said.

Oregon already has requirements for responsible AI use across its executive branch, including risk management, human accountability, transparency, and oversight. The new order focuses specifically on the state’s procurement and use of more advanced models. It directs state government to develop a policy; it does not, by itself, impose new requirements on AI companies.

The order took effect immediately. After the chief information officer submits the proposal, implementation would depend on Kotek’s approval and direction. The governor will reassess the need for the order every three months.