Operators may not operate or provide an AI companion at all unless the system has protocols in place to address three categories of user expression: (1) suicidal ideation or self-harm, (2) potential physical harm to others, and (3) potential financial harm to others. The protocol must include, at a minimum, a notification referring users to crisis service providers such as a suicide hotline or crisis text line. This is a continuous operating prerequisite — the protocol must be in place as a condition of lawfully providing the AI companion. Note that subsection (3) mentions the crisis referral obligation explicitly, but the statute structures it as applying to all three categories through the chapeau. The scope of covered harms is broader than CA SB 243, which focuses on suicidal ideation and self-harm only.

This maps the crisis referral component of § 6-63-2. Operators must maintain a protocol that includes referring users to crisis service providers (suicide hotline, crisis text line, or equivalent) when users express suicidal ideation, self-harm, physical harm to others, or financial harm to others. The crisis referral notification must be active as a condition of lawful operation. Unlike CA SB 243, this provision applies to all users — not only minors — and extends to harm-to-others scenarios beyond just self-harm and suicidal ideation.

Operators must provide a mandatory notification at the start of every AI companion interaction and at least every three hours during ongoing interactions. The notification must be delivered either verbally or in bold, capitalized text of at least 16-point type, using the prescribed language: the AI companion is a computer program, not a human being, and is unable to feel human emotion. This is an unconditional disclosure obligation — it applies regardless of whether a reasonable person would be misled. Unlike CA SB 243, where the three-hour periodic reminder applies only to known minors, this provision applies to all users. The statute also prescribes specific formatting requirements (bold, capitalized, 16-point minimum) and mandated verbatim language, which is stricter than most comparable statutes.

This provision establishes the enforcement framework for the chapter. It creates a private right of action for persons physically injured through self-harm or physically or financially harmed by another due to a violation, allowing them to sue in superior court for damages and equitable relief. It also grants the Attorney General authority to investigate, sue, and seek injunctions against noncompliant operators, including under Rhode Island's deceptive trade practices law. This creates no new substantive compliance obligation — it is purely an enforcement mechanism.

This provision directs the existing Rhode Island AI task force to continue its advisory role to the governor and general assembly on AI development and regulation. It creates no compliance obligation for AI companion operators — it is a governmental advisory mandate.