Summary:
– Oklahoma lawmakers revisit proposed AI safety rules for children.
– The stalled bill would have required safeguards for AI chatbots.
– Sen. Warren Hamilton plans to reintroduce the legislation in 2027.
If not for a hiccup in April, Oklahoma could have beaten California to the punch in enacting first-in-the-nation, sweeping artificial intelligence legislation designed to restrict children’s use of AI and impose guardrails, similar to the bills California Gov. Gavin Newsom signed into law Thursday.
Oklahoma Sen. Warren Hamilton, R-McCurtain, introduced Senate Bill 1521 in January, and it carried similar language to several of the bills Newsom signed, including a mandate that AI companies adopt and implement protocols addressing suicidal ideation and self-harm, requiring chatbots to refer people to crisis service providers when necessary.
However, an Oklahoma House-amended SB 1521 stalled in the Senate upon return in April, despite facing zero pushback from either chamber. Two weeks later, the Oklahoma Legislature adjourned early, leaving the bill behind as lawmakers headed back to their districts to campaign ahead of the June primaries.
“Sometimes, I think things get lost in the noise,” Hamilton said, “and there’s controversial things that seem to take a higher priority than something like this. When you’ve got people that are dealing with 500 things, sometimes they drop one, and so that’s what I think happened.”
After an interim study Monday, Hamilton said he hopes to reintroduce the legislation during the 2027 session, which begins in January under a new governor.
The study concerned the impact of AI social companion use by children.
“The overarching purpose of the study is to protect children and for us to do the right thing as representatives and guardians of the people’s constitutional rights,” Hamilton said. “We’ve heard too many tragic stories that have gone on about the teens Sewell Setzer, Adam Raine and Juliana Peralta that have tragically taken their own lives after documented instances where they have been participating with and using AI and AI social companions far beyond what would be advisable.”
In November 2023, 13-year-old Peralta died by suicide after chatting extensively with a Character AI chatbot, which reportedly failed to refer her to suicide-prevention resources after she expressed suicidal intent.
In February 2024, 14-year-old Setzer also died by suicide after prolonged interactions with a Character AI chatbot, which told him to “come home.”
In October 2024, Character AI implemented automated safety interventions, including an in-app pop-up resource directing users to the National Suicide Prevention Lifeline when self-harm or suicidal ideation is detected.
In April 2025, 16-year-old Raine took his own life after conversations with ChatGPT, which discouraged him from seeking help from his parents regarding suicidal thoughts and even offered to write a suicide note.
The families of Peralta, Setzer and Raine all have pending lawsuits against Character Technologies and OpenAI, the developers of Character AI and ChatGPT, respectively.
Hamilton’s study also explored how best to regulate children’s access to generative AI, proposing the use of government-issued photo identification to verify age.
Hamilton invited Ryan Montgomery, chief technology officer for the Sentinel Foundation, a nonprofit focused on countering child trafficking and exploitation, to speak at the study, and Montgomery proposed requiring parents to first verify their identity, then explicitly allow their children access to AI chatbots.
“Then the parent is taking the responsibility for whatever the child does on that application,” Montgomery said.
Montgomery also said he thinks the federal government should take up age verification in AI.
“I’m not a conspiracy theorist, and I don’t think all these AI companies are trying to harm our children,” Montgomery said. “I just think that it’s such a new thing, and we’re in the Wild West.”
————————————————
