Lifetime Citizen Portal Access — AI Briefings, Alerts & Unlimited Follows
Committee reviews H.814, a wide‑ranging bill on neurological rights and AI in health care
Loading...
Summary
A Vermont committee reviewed H.814, which would create neurological‑rights protections for neural data and regulate use of AI in health care, including consent rules for brain‑computer interfaces, disclosure requirements for generative AI, civil penalties, and a reporting requirement for the state advisory council.
A legislative committee examined H.814, a 27‑page bill that would create a new chapter on neurological rights and add rules for artificial intelligence in health and human services.
Jen Harvey, Office of Legislative Council, told members the bill begins with an intent section that "is the intent of the general assembly to protect human rights, promote equity, increase efficiency, enhance accessibility, create transparency, and guarantee accountability in health care and human services through the ethical and responsible use of artificial intelligence technology." The bill then establishes a separate Title 18 chapter recognizing individual rights to mental and neural data privacy, freedom of thought, and cognitive liberty.
The measure would generally prohibit collecting or recording neural data obtained from a brain‑computer interface without written notice and written informed consent. It would also require suppliers to obtain written consent before sharing neural data with third parties and allow individuals to revoke consent at any time; recipients must destroy neural data records within 10 days of receiving a revocation notice and must cease sharing and notify prior recipients if sharing consent is revoked.
Harvey said the bill includes limits on so‑called "consciousness bypass"—uses of brain‑computer interfaces intended to bypass an individual's conscious decision making—requiring specific written informed consent for each category of bypass use and permitting agents, guardians, or surrogates to revoke consent on behalf of an individual. The bill specifies that consent obtained using a consciousness bypass is not valid informed consent.
Enforcement provisions appear in multiple places. Harvey said a violation of the neurological‑rights chapter would be treated as an unfair or deceptive practice under Vermont's Consumer Protection Act; the transcript cites civil penalties "not more than $10,000 per violation" and gives the attorney general authority to investigate and seek remedies. Separate AI‑related provisions in the bill would also allow the attorney general to impose administrative penalties (Harvey cited up to $2,500 for each violation in one AI subsection) and to bring civil actions in superior court with remedies similar to those under the Consumer Protection Act.
The bill adds a chapter on artificial intelligence in health care with definitions for AI, generative AI, "covered‑entity generative AI," and confidential communications. It would require health‑care providers using generative AI to label written or verbal patient communications generated by AI with a clear disclaimer and provide instructions for how the patient can contact a human provider, unless a licensed human reviews and reads the communication. Violation by a licensed health care provider would fall under the jurisdiction of the Office of Professional Regulation or the Board of Medical Practice as applicable.
H.814 creates special rules for mental health chatbots. The draft defines a "mental health chatbot" as a generative AI technology that engages in interactive conversations resembling psychotherapy or other treatment and generally forbids suppliers from selling or sharing individually identifiable Vermont user data, except in narrowly defined circumstances (for example, to maintain functionality under a contractual relationship subject to HIPAA‑equivalent protections). The bill would also restrict advertising to Vermont users, require chatbots to disclose that they are AI, and impose testing, safety‑review, and reporting requirements for suppliers.
The bill provides an affirmative‑defense pathway for suppliers that file a written policy with the attorney general and can show they met development, monitoring, documentation, and best‑practice testing requirements; Harvey noted a $100 filing fee for those policies. She also warned that the affirmative defense does not prevent enforcement actions by the attorney general or licensing boards.
For health insurance uses of AI, the bill would require algorithms used for utilization review to base determinations on each covered individual's clinical record and specific clinical circumstances, prohibit reliance solely on group datasets, allow audits by state regulators, and require a licensed human to make medical‑necessity determinations.
Finally, the bill modifies the state's artificial‑intelligence advisory council by changing some appointment authorities to add education and health‑care voices, lengthening the council's sunset to June 30, 2030, and directing the council and the Division of Artificial Intelligence to produce a written report to the General Assembly by Jan. 15, 2027 with recommendations for statutory changes and pilot projects.
The committee agreed to continue detailed consideration next week and to invite witnesses, including representatives from the Division of Artificial Intelligence and the attorney general's office, to address enforcement, technical definitions, and implementation questions.
