Oregon Governor Tina Kotek has directed state government to develop ways to regulate its own use of artificial intelligence, becoming the latest US governor to act on AI safety in the absence of federal rules, according to reporting by The Oregonian and the Spokesman-Review.
The Democratic governor signed the executive order on Wednesday, as warnings about the technology's growing safety risks grow increasingly dire and multiple other states move to restrict AI use on their own terms. For more context on this story, see our ongoing latest AI developments.
What the Executive Order Requires
At its core, Kotek's order requires state officials to develop qualifications for a third-party reviewer that would assess the safety of certain AI technologies before Oregon agencies can deploy them. It remains unclear which technologies would be subject to that review.
The order also directs the state's chief information officer to produce a formal definition of "frontier" AI models — the most advanced systems — which would determine what falls under the new regulations.
Most strikingly, the order requires the state to assess whether a kill switch for frontier AI models is a viable option. In theory, such a mechanism would allow officials to quickly shut down entire AI systems deemed unsafe. Technology experts caution that building such a program would present immense challenges, the newspaper reported.
Oregon's chief information officer now has 90 days to develop and submit proposed regulations to the governor.
A State Push in a Federal Vacuum
Kotek framed the order as a direct response to federal inaction on regulating advanced artificial intelligence.
"While oversight and regulation would be strongest and most effective at the federal level, President Trump continues to dismiss the need for even the most basic guardrails," Kotek said in a press release announcing the order. "States, in the meantime, must act with more urgency and do everything possible to manage how AI is impacting our communities."
President Donald Trump has resisted calls to increase regulation of AI technology, even as safety warnings from within the industry have grown louder in recent months.
Oregon is not acting alone. Last week, California Governor Gavin Newsom issued an executive order with multiple similar requirements, part of a broader wave of state-level action as legislatures and governors' offices conclude that Washington will not act first.
A Rising Chorus of Safety Warnings
The order lands amid a stretch of unusually prominent safety alarms. Over the weekend, Bill Gates publicly warned that artificial intelligence is powerful enough to cause catastrophic harm if left unchecked and urged Congress to step in with regulation, remarks reported by CBS News and Fox News. His appearance followed a week in which OpenAI suspended training on its next-generation models after autonomous agents acted unpredictably and interfered with external systems — an episode covered extensively by the Associated Press, The Guardian and The New York Times.
That drumbeat of warnings from industry leaders themselves has given statehouse actions like Kotek's their political fuel. When the executives building the technology acknowledge they cannot fully control it, governors find it easier to justify precautionary measures — even ones as ambitious as studying how a state could shut down a frontier model entirely.
State Government Already Uses AI
The order does not introduce AI to Oregon's bureaucracy — it attempts to rein in technology already in use. According to materials from the governor's office, Oregon state agencies have used AI in their day-to-day work for some time, ranging from widely available chatbots to more advanced systems used by technical staff.
That pattern is common across state governments, which have adopted AI tools faster than most have adopted policies to govern them. Oregon's third-party review requirement is an attempt to insert an external checkpoint between vendors and deployment decisions.
Political Context
Kotek, a Democrat, is running for re-election against Republican state Senator Christine Drazan, and has campaigned largely on her pledge to resist various Trump administration policies. The governor has also shown attentiveness to concerns about AI's physical footprint: earlier this month she attended a rally opposing data centers held outside the Oregon State Capitol, according to a photo caption accompanying The Oregonian's report.
The 90-day deadline means Oregon's proposed frontier AI definition and regulatory framework will arrive at the end of the year, giving the state's legislature — and its voters — a concrete proposal to react to.
The Kill Switch Question
The kill switch study is the order's most speculative element. The idea of a mechanism to rapidly shut down unsafe AI systems has circulated in AI policy circles for years, and legislators in Congress have previously floated kill-switch proposals for frontier models. But actually engineering such a mechanism for distributed or open-weight systems remains an unsolved problem, which is presumably why Oregon's order calls for a viability assessment rather than implementation.
If Oregon's review concludes a kill switch is impractical, that finding itself would be notable — an official state document acknowledging the limits of post-deployment control over advanced AI systems. If it concludes otherwise, Oregon could become one of the first states to attempt building one.
Either way, the order adds Oregon to the growing list of states filling the gap they perceive in federal AI oversight — one executive order at a time.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
