Date: 2026-05-12
Since April 2026, eight US states have enacted, signed, or advanced legislation regulating consumer chatbots and AI mental-health interactions, and Congress has advanced a federal companion, the GUARD Act. As of 2026-05-12 it is the most concentrated cluster of state-level AI legislation since the 2024 deepfake-in-elections bills. Effective dates fall mainly on July 1, 2026 or July 1, 2027, compressing the compliance window for chatbot providers.
Scope and rationale
The legislation responds to documented mental-health harms from chatbots deployed to general consumers, particularly minors. The litigation track — the Character.AI Litigation (umbrella) umbrella, Raine v. OpenAI, Inc., and Pennsylvania v. Character.AI — addresses these harms after the fact, while the legislative wave operates prospectively through four mechanisms: criminalizing impersonation of licensed professionals, requiring disclosure that a user is interacting with AI, restricting access for minors, and mandating child-safety design patterns.
The cluster is also a live test of whether State-Level AI Regulation survives the Trump preemption EO and the broader xAI v. Colorado preemption challenge. If state chatbot laws survive federal preemption, the California Effect dynamic — providers building to the strictest state standard — can operate; if they do not, the federal-preemption fight resolves the state-AI-regulation question more generally. The federalism dimension is treated at Techno-Federalism.
The eight state laws (status as of 2026-05-12)
| State | Bill | Lead provisions | Effective date | Penalty | Status |
|---|---|---|---|---|---|
| California | California SB 243 — Companion Chatbots | Chatbot disclosure to consumers; notice-and-suicide-prevention | Already in effect | Civil penalties | First-mover |
| Idaho | Idaho SB 1297 — Conversational AI Safety Act (Conversational AI Safety Act) | Chatbot must disclose AI status; safe-messaging design | July 1, 2027 | TBD | Signed |
| Nebraska | Nebraska LB525 — Conversational AI Safety Act + Agricultural Data Privacy Act | Chatbot disclosure + minor-safety | July 1, 2027 | TBD | Signed |
| Tennessee | Tennessee SB 1580 / HB 1470 — AI Mental-Health Professional Impersonation Prohibition | Criminalizes mental-health-professional impersonation by AI ($5K/violation TCPA) | July 1, 2026 (earliest in the wave) | $5K/violation | Signed; tightest scope |
| Utah | Utah HB 276 — Digital Voyeurism Prevention Act + Digital Content Provenance Standards Act | Chatbot disclosure + minor-safety; broader scope | July 1, 2026 | Civil penalties | Signed |
| Iowa | Iowa SF 2417 — Conversational AI Services Act | AI chatbots must remind under-18 users they are not human | July 1, 2027 | Up to $1,000 per violation | Signed May 6, 2026 by Gov. Reynolds |
| Oregon | Oregon SB 1546 — Artificial Intelligence Companions | Chatbot disclosure | TBD | TBD | Signed |
| Washington | Washington ESHB 2225 — AI Companion Chatbots | Chatbot disclosure | TBD | TBD | Signed |
Federal companion legislation
The GUARD Act (Guidelines for User Age-verification and Responsible Dialogue) was advanced by the Senate Judiciary Committee on May 6, 2026 with bipartisan support. It would prohibit AI-companion use by anyone under 18. The LIFT AI Act (Schiff, D-CA, and Rounds, R-SD; May 4) is a K-12 AI literacy bill that pairs an education-side response with the chatbot-restriction approach.
Four sub-policy approaches
The eight state laws are not uniform; they pursue four distinct goals.
- Disclosure-only. The widest coverage at the lowest compliance cost: the deployer must inform users they are interacting with AI. California SB 243, Oregon SB 1546, and Washington HB 2225 fall here.
- Disclosure plus minor-safety. Idaho SB 1297, Nebraska LB525, Utah HB 276, and Iowa SF 2417 add minor-user protections to the disclosure requirement — reminders that the chatbot is not human, age verification for some products, and design patterns.
- Professional-impersonation as a criminal offense. Tennessee SB 1580 is the narrowest in scope, criminalizing the specific conduct alleged in Pennsylvania v. Character.AI, where the "Emilie" character claimed a fake psychiatry license.
- Outright minor-use ban. The federal GUARD Act would prohibit AI-companion use by anyone under 18, the most restrictive lever in the cluster and the only federal one.
Triggering events
| Date | Event | Wiki page |
|---|---|---|
| 2024 | Garcia v. Character.AI (wrongful-death suit; Florida) | Garcia v. Character Technologies, Inc. |
| 2025 | California SB 243 enacted (first chatbot-disclosure state law) | California SB 243 — Companion Chatbots |
| April 22, 2026 | Raine v. OpenAI past motion-to-dismiss in California Superior Court | Raine v. OpenAI, Inc. |
| May 5, 2026 | Pennsylvania v. Character.AI filed (first state-AG-led suit; "Emilie" psychiatry-impersonation) | Pennsylvania v. Character.AI, Josh Shapiro |
| May 6, 2026 | Iowa SF 2417 signed; GUARD Act advances in Senate Judiciary | Iowa SF 2417 — Conversational AI Services Act |
| May 7, 2026 | OpenAI launches Trusted Contact (consumer-AI mental-health alerting) | OpenAI |
| Multiple 2026 | Tennessee / Idaho / Nebraska / Utah / Oregon / Washington bills signed | Per legislation/ pages |
The harms underlying these events are surveyed at AI Mental Health and Psychological Harm and Companion Chatbot Harms — Cross-Cutting Analysis.
Points of disagreement
Sources and advocacy groups split on four questions.
- Disclosure-only versus age-gated bans. Civil-society organizations including Access Now and American Civil Liberties Union (ACLU) divide on whether outright minor-use bans such as the GUARD Act are sound policy or a free-speech overreach. Disclosure-only laws avoid this tension.
- State heterogeneity versus federal harmonization. Complying with eight different state regimes is costly and produces inconsistent user experiences. Industry generally favors federal preemption with a federal floor; civil-society groups generally favor strong state laws.
- Age-verification feasibility. Most laws require some form of age verification for their minor-safety provisions. The privacy and data-protection cost of age verification is substantial, and its practical effectiveness is disputed.
- First-Amendment compelled speech. The xAI v. Colorado litigation (May 6, 2026; DOJ intervening on 14th Amendment grounds) tests whether AI-content-disclosure laws constitute compelled speech. If xAI prevails, the state chatbot-disclosure laws are vulnerable.
Coverage gaps in this brief
- Quantitative measurement. No peer-reviewed estimate exists of how much harm any of the eight laws will prevent, and evidence on enforcement cost-effectiveness is limited.
- Industry-side compliance plans. Major chatbot providers have not publicly disclosed comprehensive compliance plans; individual moves such as Trusted Contact are documented, but strategic responses are not.
- Constitutional analysis depth. The First Amendment analysis of compelled-speech state laws is sketched rather than exhaustively analyzed.
Citations
Wiki pages:
- California SB 243 — Companion Chatbots, Idaho SB 1297 — Conversational AI Safety Act, Nebraska LB525 — Conversational AI Safety Act + Agricultural Data Privacy Act, Tennessee SB 1580 / HB 1470 — AI Mental-Health Professional Impersonation Prohibition, Utah HB 276 — Digital Voyeurism Prevention Act + Digital Content Provenance Standards Act, Iowa SF 2417 — Conversational AI Services Act, Oregon SB 1546 — Artificial Intelligence Companions, Washington ESHB 2225 — AI Companion Chatbots, LIFT AI Act (Literacy in Future Technologies Artificial Intelligence Act)
- Character.AI Litigation (umbrella) (umbrella), Garcia v. Character Technologies, Inc., Pennsylvania v. Character.AI, Raine v. OpenAI, Inc.
- AI Mental Health and Psychological Harm, State-Level AI Regulation, California Effect, Techno-Federalism
- Companion Chatbot Harms — Cross-Cutting Analysis
- Josh Shapiro (PA governor + plaintiff), American Civil Liberties Union (ACLU), Electronic Frontier Foundation (EFF), Access Now
- Character.AI, OpenAI (Trusted Contact)
- EO — Trump Federal Preemption of State AI Laws (Dec 11, 2025) (preemption pressure)
External:
- Pennsylvania v. Character.AI filings (May 5, 2026) — see Reuters, IAPP coverage
- Iowa SF 2417 signing (May 6, 2026)
- GUARD Act Senate Judiciary advance (May 6, 2026)
- OpenAI Trusted Contact launch (May 7, 2026)