Oregon Orders AI Safety Reviews for State Contracts

Oregon Governor Tina Kotek signed an executive order on Wednesday. The directive establishes new safety standards for how state agencies buy and use artificial intelligence tools.

The order aims to prevent risky technology from entering government systems. It directs state officials to create rules for reviewing AI safety before adoption.

New rules for AI safety

The order tells the state chief information officer to develop specific criteria. These standards will guide third-party reviews of AI safety. The goal is to ensure tools work as intended without causing harm.

Kotek stated that federal regulation is currently lacking. She argues that states must act independently to protect their communities. This move fills a gap in national oversight.

Assessing emergency shutdown options

The directive also requires an assessment of a specific safety feature. Officials must evaluate the viability of a kill switch. This is a control that can instantly stop an AI model from operating.

Kotek emphasized that companies cannot be the sole guardians of public safety. The state will not rely on voluntary industry standards. External checks are necessary to manage risks effectively.

Timeline and future review

The order takes effect immediately upon signing. State officials have ninety days to submit an implementation proposal. This deadline allows for a rapid response to emerging risks.

Kotek will reassess the necessity of the order every three months. This periodic review ensures the rules remain relevant. The order stays in effect until the governor terminates it. KATU reported that this approach balances urgency with legislative oversight.