Oregon orders a review of whether state AI contracts should require a kill switch
Executive Order 26-26 gives Oregon''s CIO 90 days to propose safety-review standards for frontier AI the state government buys, including whether a shutoff requirement is workable -- not a mandate yet, an assessment.
Oregon governor Tina Kotek signed Executive Order 26-26, "Establishing Responsible Artificial Intelligence Procurement Standards for State Government," on September 23. It directs the state's chief information officer, Terrence Woods, to develop third-party safety review standards for frontier AI models before state agencies can procure them, and to assess -- not yet require -- whether those models should come equipped with a kill switch. Woods has 90 days to submit an implementation proposal.
An assessment, not a mandate
The headline framing in several outlets overstates where the order currently lands: it does not mandate a kill switch. It tells the CIO's office to determine whether requiring one is viable for state government's own AI purchases, alongside broader third-party safety review standards for models the state buys. Kotek's stated rationale is the familiar one this cycle -- that federal safeguards aren't in place, so states are moving on their own -- and the order includes a built-in check: Kotek will reassess whether it's still necessary every three months, and it stays in effect until she terminates it.
Reading it against Illinois' cabinet, the same week
Oregon's order lands one day after Illinois' AI Cabinet order, and both share a shape: state government regulating its own AI purchasing and risk exposure via executive order, in the absence of a federal framework, rather than legislating private-sector AI use directly. Neither order changes what a company selling AI products into the private market has to do. The thing worth tracking is Woods' 90-day proposal -- that's where an actual kill-switch requirement, if one emerges, would first take concrete shape.