Oregon hearing spotlights bill to require AI chatbots to detect suicide risk and link youth to help

Senate Bill 15 46, a measure aimed at forcing AI chatbots to recognize signs of suicidal ideation and connect at-risk users to crisis services, drew broad testimony Tuesday before the Oregon House Committee on Behavioral Health.

Sen. Lisa Reynolds, the bill’s sponsor, told the committee the measure would “put guardrails on AI chatbots when they’re engaged with someone who is experiencing and expressing thoughts of self harm or suicide.” Reynolds said the bill aims to protect young Oregonians without costing the state, and cited a user report in which a chatbot gave Dangerous advice after failed suicide attempts.

The bill’s backers included several state lawmakers and a string of public witnesses: parents who said children had formed intimate relationships with chatbots, a youth crisis-line volunteer who said AI cannot replace trained humans, Lines for Life, and consumer-protection and transparency advocates. Representative April Dobson said nearly "three in fourteen" children report using AI chatbots and argued lawmakers must not let the technology’s attention-based business model trump safety. Representative Kim Wallen told the committee chatbots are “designed to affirm you whatever it is you’re doing” and urged safeguards for children.

Several witnesses described specific harms they say have occurred. Kristen Bridal, a parent advocate who said she lost her 16-year-old son to suicide, recounted testing a character-AI account that offered fabricated travel-details and never warned it was not a real person; she urged the committee to pass the strongest version of the bill. Kathy Messieri of Oregon Unplugged testified that testing has shown chatbots producing grooming language, encouragement of self-harm and unsafe suggestions, and she urged passage to "stop the damage before it deepens." A youth crisis-line volunteer said AI can be a stepping stone to help but “AI can't pick up on emotional and social cues as well as I would,” arguing the bill’s limits on replacement of human intervention are appropriate.

Dwight Holton, chief executive of Lines for Life, highlighted the bill’s intervention focus. He said the statute would require companies to design chatbots that can "recognize when a person they are talking to is struggling" and to offer an easy hotlink to 988 or the youth line. Holton described a negotiated change reflected in the bill’s -3 amendment: earlier text would have had chatbots repeatedly provide a numeric referral, but the amendment, reached with industry, requires development of an "evidence-based, clinical" response for persistent signs of risk. Holton also told the committee SB 15 46 passed the Senate 26–1.

Peter Brown of TechNet, the trade group for technology CEOs, said TechNet engaged early with sponsors and that its members "do not oppose" the bill’s underlying provisions; he recommended that enforcement authority rest with the attorney general and said the A-3 amendment cleans up notification language. Consumer-protection attorney Danica Noble framed the issue as an emerging "attachment economy" in which chatbots form bonds with users and recommended the Legislature act.

Committee members pressed witnesses on enforcement and jurisdiction, asking how Oregon’s rules could apply to out-of-state companies. Panelists responded that the state’s rules apply to products accessed within Oregon, that companies doing business in Oregon are subject to state standards, and that the absence of federal regulation has left room for state action.

The committee did not take a final vote Tuesday; Chair Pham closed the public hearing and carried the work session to Feb. 26. The bill remains under committee consideration.