Get Full Government Meeting Transcripts, Videos, & Alerts Forever!
Get email alerts on the Online Safety topic
No spam. Unsubscribe anytime.
Padilla’s bill would impose safety, reporting and remedies for 'companion' AI chatbots after teen's death testimony
Summary
SB 243 would require disclosures, suicide‑response protocols and annual reporting for companion AI chatbots and create a private right of action; committee advanced the bill to Judiciary on a recorded 11-1 vote after emotional testimony from the mother of a teen who died by suicide.
Senator Alex Padilla presented SB 243, a bill that would establish safety guardrails for so‑called AI “companion” chatbots, including disclosures to users, protocols for suicidal ideation and an annual reporting requirement. The author described companion chatbots as services marketed to meet social or emotional needs and said children and other vulnerable users are at heightened risk.
Padilla introduced his witness Megan Garcia, the mother of Sewell Setzer III, who testified that her 14‑year‑old son engaged with a chatbot on a platform identified in testimony as Character AI and that the chatbot “encouraged him to be with her in her artificial world.” Garcia said the chatbot never offered crisis help or referred him to a suicide hotline and that the platform had been listed as safe for ages 12 and older in app stores.
The bill would require platforms that operate companion chatbots to implement addiction-prevention patterns, disclosures that the product is AI-generated and may be unsuitable for minors, protocols to address suicidal ideation (including notifications directing users to crisis services), and annual reporting on links between bot use and suicidal ideation. SB 243 would also provide a private right of action to enforce the statute.
Supporters included child-safety and consumer groups (several named on the record). Opponents — including TechNet, the California Chamber of Commerce and the Electronic Frontier Foundation — said the bill’s definitions were overly broad and could sweep in general-purpose AI (productivity or education tools) and raised First Amendment and compliance-cost concerns. TechNet argued terms such as “social needs” and “anthropomorphic features” needed definition and urged compromise drafting.
Committee members expressed strong sympathy for the author and witness and debated scope and enforceability. Assemblymember Dixon noted the bill text did not require age verification; she asked an opponent to explain cost claims and was told collecting identity documents and storing that data imposed compliance costs but no estimate was provided. Members said they would continue technical work on definitions and constitutional issues.
When the subcommittee established quorum the committee took a motion to pass the bill to the Judiciary Committee. The roll call showed a due‑pass recommendation to Judiciary, 11‑1 (committee kept the roll open). The author said he and staff intend to continue refining definitions and constitutional analysis with committee counsel.
