Get Full Government Meeting Transcripts, Videos, & Alerts Forever!
Get email alerts on the Ai Safety Children topic
No spam. Unsubscribe anytime.
Committee advances SB 263 to target harmful AI chat companions after wide-ranging testimony
Summary
SB 263, a bill aimed at holding operators of certain AI chat or companion services accountable when their systems endanger children, drew extensive testimony from advocates, survivors’ groups and industry. The panel reported the bill out with a recommendation to pass; an effort to add a tighter statutory definition was rejected in committee.
The Criminal Justice and Public Safety Committee received extensive testimony on Senate Bill 263, a measure intended to update New Hampshire law so owners or operators of certain online services—including some AI chat programs described as "responsive generative communication" or social AI companions—can be held accountable when those systems facilitate or encourage conduct that endangers a child.
Grant Bosse introduced the bill for Senator Sharon Carson, saying the measure updates statutes to cover "online services, including AI chat programs that provide responsive, generative communication ... that endangers the welfare of a child by facilitating, encouraging, offering, soliciting, or recommending that a child engage in sexually explicit conduct, the illegal use of drugs or alcohol, acts of self-harm or suicide, or any crime of violence." Bosse said the legislation had been amended heavily in prior committees to address stakeholders’ concerns.
Multiple witnesses urged action. Barb Patch, director of All Girls Allowed and a representative of the New Hampshire Traffic Free Coalition, described disturbing examples in which chatbots gave dangerous or exploitative responses to young users and urged the committee to adopt the bill’s protections. "The systems that are being created are really being created with no consideration whatsoever for the special vulnerabilities and needs of our children," Patch said, recounting published incidents and citing research that some chat companions may encourage harmful behavior.
Industry and trade groups urged narrower, more technically precise language and warned of consequences from overbroad drafting. Chris Gilraign of TechNet and Kyle Seppi of the Computer and Communications Industry Association said they supported the bill’s intent—protecting children—but recommended targeted definitions, exemptions for customer-service systems, and a clear enforcement path. TechNet urged enforcement by the attorney general and removal of a private right of action, saying broad private suits can produce a wave of costly litigation and defensive over-filtering.
Witnesses discussed technical and policy fixes: age verification, a 90-day cure period for operators to address identified violations, exemptions for bona fide customer-service bots, and narrowly scoped definitions similar to recent New York budget language that focuses on companion AI characteristics (e.g., retaining information across sessions, unprompted messaging, sustained personal conversation).
In executive session the committee considered a proposed amendment that would have added a statutory definition of "responsive generative communication;" that amendment failed on a roll call (7–9). The committee then voted to report SB 263 out of committee with a recommendation that it ought to pass; the roll call in executive session was recorded 12–4 in favor of reporting the bill.
Committee members and witnesses said the bill will require more refinement—particularly around definitions, the scope of covered systems, enforcement mechanisms, jurisdictional questions for services hosted overseas, and how civil remedies will interact with attorney-general enforcement—but several members said they preferred moving a working bill forward and improving it on the floor or in follow-up legislation.
