Great job, Tina!
The first bill (2026-S 2195Aaa, 2026-H 7350Aaa) requires developers to incorporate protocols to protect vulnerable users who may express suicidal ideation, desires for self-harm or desires to physically harm others while interacting with a chatbot.
The systems would also be required to refer users to a crisis service provider as soon as it detects any of these expressions, and developers would have to submit annual reports to the Attorney General’s office showing the number of safety protocol activations and related metrics.
“AI chatbots are a new kind of technology that provide an
experience that to many users feels like speaking to a friend or a therapist.
But chatbots aren’t human and they do not live up to the professional standards
of a trained therapist,” said Representative Spears (D-Dist. 36, Charlestown,
New Shoreham, South Kingstown, Westerly). “This legislation puts up necessary
guardrails to protect the users of chatbots, because, as we have seen in too
many tragic instances, the products are not currently equipped to deal with
users experiencing a crisis in a safe and responsible manner. Whether at home
or in a clinical setting, human oversight and thoughtful safety regulation are
necessary to keep Rhode Islanders safe from a growing and untested technology.”
Multiple parents of teenagers who died by suicide testified before Congress last year that their children had extensive conversations with AI companions that failed to refer them to crisis resources when they expressed suicidal desires or attempt to discourage them. In one case the chatbot even offered to help draft a suicide note.
The act would also require the chatbot to clearly and
conspicuously notify its users that they are not communicating with a human at
the beginning of a session and at least once every three hours during an
ongoing session.
The second bill (2026-S 2197Aaa, 2026-H 7349A) would establish guidelines regarding the
permitted and restricted uses of AI technology in mental healthcare, in an
effort to safeguard patient safety and privacy. It would require the patient or
their parent or guardian’s fully informed consent before AI companion models
could be used to assist in therapy or psychotherapy sessions, and ensure that
licensed mental health professionals, not AI, are always the ones making
therapeutic decisions, determining therapeutic recommendations and creating
treatment plans.
It would also ensure that AI tools used in mental healthcare
— including those assisting in administrative tasks, notetaking, supplementary
support and therapeutic communication — meet the same confidentiality standards
that currently apply to all other records and communications kept by licensed
mental health professionals.
In written testimony submitted in support of the bill,
Family Service of Rhode Island (FSRI), which provides behavioral health and
social services across the state, wrote “an over-dependence on AI to guide
clinical treatment can result in errant decision-making and harmful outcomes in
client care. We fully support the implementation of regulations regarding the
use of AI in mental healthcare treatments to ensure all clients are properly
informed of the details and the specific purpose for the use of AI in their
treatment.”
Beth Bixby, CEO of Tides Family Services, also wrote in
support of the bill, saying “The bill appropriately limits AI from engaging in
therapeutic communication or making clinical decisions. Maintaining this
boundary is essential—especially for youth involved in juvenile justice and
child welfare, where trust, clinical judgment and family engagement are central
to outcomes.”
The bill would also prevent companies from using AI
companion models as an alternative to licensed therapists by prohibiting
companies and individuals from offering or advertising therapy or psychotherapy
services unless those services are conducted by a licensed professional.
Both bills now go to the governor for his consideration.
