A Missouri Senate committee heard testimony Wednesday on three bills that would regulate artificial intelligence.
AI is rapidly expanding, and its capabilities are unknown, said Sen. Mike Moon, R-Ash Grove, who said he has experimented with an AI companion, Pi.
"It never told me that I was wrong," Moon said. "It appears that these systems always are affirming us as individuals and what we want to hear or do."
Moon is sponsoring one of the bills heard by the committee. It would declare that AI is a nonsentient entity that cannot own property, be recognized as a person or spouse or serve as a manager or director of a company.
Under the bill, any harm caused by AI would be the responsibility of the owner or user, and developers would be required to develop safety mechanisms.
"We need to set some framework of guidelines, so that we can know what the limitations should be and do our best to avoid harm and allow the technology to expand within safe guidelines," Moon said.
He said the bill does not violate the wishes of President Donald Trump, who in December launched a task force to sue states over their AI-related laws.
People spoke both in favor of and in opposition to Moon's legislation.
"To have some guardrails, to have some definitions, I think, are really important," said Brittany Pratt, a clinical psychologist who testified that she's witnessed people using AI for friendship, mental health support and financial advice.
Jared Hankinson, with the Missouri Chamber of Commerce, said the legislation contains unclear definitions that could invite litigation for the businesses that develop AI tools.
"It does serve as a deterrent from them investing in the state of Missouri, bringing those dollars and some of those jobs to the state," Hankinson said.
Similar legislation was heard in a House committee last month but has yet to make it to the House floor.
AI and mental health
Another bill heard by the committee would prevent developers from representing AI as a service that can provide therapy.
"A person cannot falsely advertise that they are a licensed mental health professional," said Rep. Patty Lewis, D-Kansas City, the bill's sponsor. "The same logic should apply to AI."
Lewis cited a recent study that found one in eight teenagers has turned to large language models for mental health advice.
Chatbots can be dangerous for that purpose because they have been known to hallucinate, Lewis said, describing a speech she prompted ChatGPT to write about her that included she was a "proud Kansan."
"I am a subject matter expert on myself, and I can tell you I am not a proud Kansan," Lewis said.
Nearly 15 people testified in support, including some representing the Missouri chapter of the American Academy of Pediatrics, the Missouri Mental Health Counselors Association and the Missouri Psychological Association.
Pratt said she experienced the potential harms of AI firsthand when her daughter secretly used a home smart assistant for advice about bullying – and began to resist advice from her parents before they were able to put guardrails in place.
"Two days is how long it took an 8-year-old to trust one of these AI chatbots over a human that they had a long-standing relationship with," Pratt said.
No one testified in opposition to Lewis' bill.
AI watermarks
A third piece of legislation, sponsored by Sen. Brad Hudson, R-Cape Fair, would require those creating AI-generated content to clearly label it via watermarks or disclaimers.
The bill would also require anyone who uses AI to depict a real person to obtain their consent, unless the content is used for parody or satire.
Hudson said he crafted the legislation after actor Barry Williams – known for his role as Greg Brady in "The Brady Bunch" – and his wife raised concerns.
"They are constituents of mine and have had to endure horrible deep fakes created of them and shared on the internet," Hudson said. "Regardless of fame, no one should have to endure that invasion of privacy."
Hankinson said placing watermarks on AI-generated content would be too difficult for many businesses.
"There is, for lack of a better term, a boatload of artificially generated content out there, and it would be a very significant cost burden," Hankison said.
An advertising firm and video game developer also testified in opposition.
Hudson said he is willing to discuss the bill with anyone who is interested.
"I want this to be a work in progress," Hudson said. "This is the beginning of a framework."
SBs 859, 1444 and 1324 will need to be approved by the General Laws committee before they reach the Senate floor.
Copyright 2026 St. Louis Public Radio
