Arizona HB 2311 (artificial intelligence service; disclosures; requirements)

House Bill 2311 of the Arizona Legislature's 2026 session would have added a new chapter 8, "Artificial Intelligence," to title 18 of the Arizona Revised Statutes, placing disclosure, content-restriction, minor-protection, crisis-response and age-assurance privacy duties on operators of publicly available conversational AI services. Enforcement would have rested solely with the attorney general, with civil penalties of $1,000 per violation up to $500,000 per operator and no private right of action. The Legislature passed the bill on June 12, 2026, and Governor Katie Hobbs vetoed it on June 19, 2026, writing that it limited damages to families, barred families from suing and "ties the state's hands in bringing actions in an unprecedented manner" (HB 2311 Veto Letter — Gov. Katie Hobbs (Arizona, June 19, 2026)).

Sources for the table: bill-status record and House summary (Source: apps.azleg.gov; azleg.gov).

Status and legislative history

Rep. Tony Rivero introduced the bill on January 14, 2026 (Source: azleg.gov; apps.azleg.gov). The House Committee on Artificial Intelligence & Innovation recommended it do pass amended, 7–0, on February 12, and the House passed it on third reading on February 24 by 43–13. In the Senate, the Committee on Appropriations, Transportation and Technology recommended it do pass on March 17 by 7–2 with one member not voting. The bill was amended twice on the Senate floor by Sen. David C. Farnsworth in the Additional Committee of the Whole. After the first amendment the Senate passed the bill on June 9 by 28–1; it reconsidered that vote on June 10, adopted a second amendment on June 11 and passed the bill again by 28–0 with two not voting. The House concurred in the Senate amendments and passed the bill on final reading on June 12 by 35–20 with five not voting. It was transmitted to the Governor on June 13 and vetoed on June 19, 2026 (Source: apps.azleg.gov; azleg.gov).

The vetoed text is the Senate Engrossed version of June 11, 2026 as amended by the second Farnsworth floor amendment, which the Legislature adopted with reference to that engrossed bill. The Senate Engrossed version posted on the Legislature's website does not incorporate that amendment, and no later engrossed or enrolled version was published (Arizona HB 2311 — Bill Text (Senate Engrossed with Farnsworth Floor Amendment #2, 2026)).

Scope and definitions

A "conversational AI service" is an artificial intelligence software application, web interface or computer program that is accessible to the general public and that primarily simulates human conversation and interaction through textual, visual or aural communications. The definition excludes nine categories: products primarily designed and marketed for developers or researchers; a feature within another application, web interface or program that is not itself a conversational AI service; products designed to provide outputs on a narrow and discrete topic; products primarily designed and marketed for commercial use by business entities; products that incorporate a speaker and voice command interface or a text interface and act as a virtual assistant for a consumer electronic device; products used by a business entity solely for internal purposes; products used by a business entity or by a state or local government agency solely for customer service or information about its own services or products; and products used solely for commerce-related or transactional assistance (Arizona HB 2311 — Bill Text (Senate Engrossed with Farnsworth Floor Amendment #2, 2026)).

An "operator" is a person that makes a conversational AI service available to the public, excluding a mobile application store or search engine solely because it provides access to one. A "minor" is an individual whom the operator has "actual knowledge or reasonable certainty" is under eighteen. An "account holder" is an individual who has or opens an account or profile to use a conversational AI service, and most duties are expressed as owed to account holders. The bill also defines a "digital identification system" (a process using government-issued identification, facial recognition, facial age-estimation or age-classification technology, biometrics or another uniquely identifying credential) and an "age-assurance method" (any technical or administrative mechanism used solely to determine whether an account holder is a minor) (Arizona HB 2311 — Bill Text (Senate Engrossed with Farnsworth Floor Amendment #2, 2026)).

Key provisions

Disclosure

Each operator would have had to disclose clearly and conspicuously to each account holder that the account holder is interacting with a conversational AI service, either as a persistent visible disclaimer or at the beginning of each session and at least every three hours in a continuous interaction (§ 18-802(A)) (Arizona HB 2311 — Bill Text (Senate Engrossed with Farnsworth Floor Amendment #2, 2026)).

Content and engagement restrictions

An operator that knows an account holder is a minor could not provide points or similar rewards at unpredictable intervals with intent to encourage increased engagement (§ 18-802(B)). For all account holders, operators would have had to institute reasonable measures to prevent the service from producing visual material of sexual conduct, generating direct statements that the account holder should engage in sexual conduct, or generating statements that sexually objectify the account holder (§ 18-802(C)). For minor account holders, operators would have had to institute reasonable measures to prevent statements that would lead a reasonable person to believe they are interacting with a human, including explicit claims that the service is sentient or human, statements simulating emotional dependence, statements simulating romantic or sexual innuendos, and role-playing of adult–minor romantic relationships (§ 18-802(D)) (Arizona HB 2311 — Bill Text (Senate Engrossed with Farnsworth Floor Amendment #2, 2026)).

Parental tools, crisis protocol and professional representations

Operators would have had to offer privacy and account-settings tools to minor account holders and, for a minor under thirteen, to the parent or guardian, with related tools for parents of older minors "as appropriate based on relevant risks"; the tools may be local to the device or account and may not require a digital identification system (§ 18-802(E), as amended). Operators would have had to adopt a protocol for responding to prompts about suicidal ideation or self-harm, including reasonable efforts to refer the account holder to crisis service providers such as a suicide hotline or crisis text line, and institute reasonable measures preventing statements that encourage, instruct or glorify suicide or self-harm (§ 18-802(F)). An operator could not knowingly and intentionally cause or program the service to represent that it is designed to provide professional mental or behavioral health care (§ 18-802(G)) (Arizona HB 2311 — Bill Text (Senate Engrossed with Farnsworth Floor Amendment #2, 2026)).

Age assurance and data protection

Unless federal law expressly requires it, an operator could not condition access to a conversational AI service on the use of a digital identification system for age verification; an operator that voluntarily offers one must provide a privacy-preserving alternative with equivalent access. An age-assurance method may collect only the minimum data reasonably necessary, which may not be repurposed for advertising, profiling or unrelated analytics and must be deleted or irreversibly de-identified once the compliance purpose is satisfied. The subsection does not require account creation or prohibit anonymous or pseudonymous use (§ 18-802(H), as amended). Compliance data would have been subject to reasonable safeguards and the state's breach-notification statute, § 18-552, and barred from targeted advertising, behavioral profiling or secondary monetization (§ 18-802(I)). A governmental entity could compel disclosure of that data only under a warrant issued on probable cause, with notice to the account holder within seventy-two hours unless the court order prohibits it (§ 18-802(J)), and an operator could not transfer or make available to a governmental entity any data, model, analytics or profile derived from compliance except under such a warrant (§ 18-802(K)). By April 1 each year, an operator collecting such data would have had to certify under penalty of perjury that it had been destroyed or de-identified, and every operator would have had to publish an aggregate report identifying its age-assurance methods and whether a digital identification system and alternatives are offered (§ 18-802(L)) (Arizona HB 2311 — Bill Text (Senate Engrossed with Farnsworth Floor Amendment #2, 2026)).

Enforcement and penalties

An operator violating the chapter would have been subject to an injunction and liable for the greater of actual damages or civil penalties of $1,000 per violation, not to exceed $500,000 per operator (§ 18-802(M)). A violation is punishable by a civil penalty "to be sought by the attorney general only," and the section "does not create a private right of action to enforce this section or to support a private right of action under any other law." The attorney general could not adopt a rule, guidance or enforcement action, or enter into a settlement agreement, that expands a requirement beyond those expressly included in the section, "including a requirement for identity verification or bulk data reporting" (§ 18-802(N)) (Arizona HB 2311 — Bill Text (Senate Engrossed with Farnsworth Floor Amendment #2, 2026)). Idaho's SB 1297 and Nebraska's LB525, both enacted in 2026, use the same $1,000-per-violation and $500,000-per-operator figures, reserve enforcement to the attorney general and exclude a private right of action (Source: Raw Sources/Idaho Conversational AI Safety Act (SB 1297).md; Raw Sources/Nebraska Conversational AI Safety Act (LB525).md).

Rules of construction

The section would not have created liability for the developer of an AI model for a violation by a conversational AI service that a third-party operator makes available (§ 18-802(O)(1)). It was to be construed in the least intrusive manner consistent with article II, section 8 of the Arizona Constitution and could not be used to implement a system that tracks any of an account holder's online activity; it did not authorize regulation of lawful political, religious or other protected speech; it could not be used as a predicate to require digital identification for general internet, device or online activity unrelated to a conversational AI service; it could not be construed to require or authorize an operating-system provider, application store, internet service provider or device manufacturer to implement age assurance or identity authentication on an operator's behalf; and it could not be cited or used to create or enforce a social scoring system, a social credit system or any system that rates, ranks or restricts access to a service based on an individual's behaviors, expressions or associations (§ 18-802(O)(2)–(6), as amended) (Arizona HB 2311 — Bill Text (Senate Engrossed with Farnsworth Floor Amendment #2, 2026)).

Changes during passage

The bill as introduced defined an operator as a person that "develops and makes available" a conversational AI service; the House version narrowed this to one that "makes available" a service, and added exclusions for business customer-service and commerce-related products. In the introduced and House versions, the disclosure duty and the sexual-content measures applied only to minor account holders, and a separate subsection required disclosure where a reasonable person would be misled into believing they were interacting with a human (Source: azleg.gov; azleg.gov).

The first Senate floor amendment (June 9) extended disclosure and the sexual-content measures to every account holder, added the duty to prevent statements encouraging or glorifying suicide or self-harm, added the age-assurance and data-protection provisions of subsections H through L, added the limit on the attorney general's rules, guidance, enforcement actions and settlements, added the rules of construction on speech, digital identification and device-level age assurance, reworded the voice-assistant exclusion, added the government customer-service exclusion and removed the definition of "visual depiction" (Source: azleg.gov). The second amendment (June 11) replaced the bar on requiring a digital identification system "solely to determine whether an account holder is a minor" with the bar on conditioning access on one "for age verification," changed the tracking limit from "all" to "any" of an account holder's online activity, and added the social-scoring clause (Arizona HB 2311 — Bill Text (Senate Engrossed with Farnsworth Floor Amendment #2, 2026)). The $500,000 per-operator ceiling, attorney-general-only enforcement and the bar on a private right of action were present in every version (Arizona HB 2311 — Bill Text (Senate Engrossed with Farnsworth Floor Amendment #2, 2026); Source: azleg.gov).

Veto

Hobbs's letter to House Speaker Steve Montenegro states: "I will not protect big technology companies and AI chat bots more than children." It objects that the bill "limits damages to families to what amounts to a drop in the bucket for large corporations, prohibits families from bringing their own lawsuits, and ties the state's hands in bringing actions in an unprecedented manner," and says Hobbs looks forward to working with the Legislature and parents "on how to prevent harm to our youth without simultaneously shielding technology companies from meaningful liability." The letter does not cite subsections; the three objections correspond to subsections (M) and (N) above. It does not address the disclosure, content, minor-protection, crisis or age-assurance provisions (HB 2311 Veto Letter — Gov. Katie Hobbs (Arizona, June 19, 2026)). HB 2311 was one of three bills with "artificial intelligence" in their short titles that Hobbs vetoed on June 19, 2026, alongside HB 2592 (state agencies; rules) and HB 4005 (course; public schools) (Source: azgovernor.gov).

Comparison with Colorado HB 26-1263

Colorado's Chatbot Safety Act, signed May 29, 2026, addresses the same category of service with a near-identical core definition: an AI system accessible to the general public that primarily simulates human conversation and interaction. Colorado lists twelve exclusions against Arizona's nine; both exclude developer and research tools, features embedded in other applications, device voice assistants and commerce or transactional assistance, and both exempt application stores and search engines from the operator definition (Colorado HB 26-1263 (Chatbot Safety Act, Enrolled Act); Arizona HB 2311 — Bill Text (Senate Engrossed with Farnsworth Floor Amendment #2, 2026)).

The two differ on age determination and remedies. Colorado requires operators to estimate users' ages by commercially reasonable or generally accepted methods and treats an estimate as knowledge of minority (Colorado HB 26-1263 (Chatbot Safety Act, Enrolled Act)); HB 2311 defined a minor by the operator's actual knowledge or reasonable certainty and barred conditioning access on a digital identification system for age verification. Both lack a private right of action and route enforcement through the attorney general: Colorado's act contains no penalty provision and relies on the Part 17 mechanism of the Colorado Consumer Protection Act (Colorado SB 26-189 (Signed Act, May 14 2026)), while HB 2311 set its own civil penalties with a per-operator ceiling and limited the attorney general's rules, guidance and settlements.

Drafting context

NPR reported on September 18, 2026 that chatbot bills introduced in at least ten states in 2026 carried similar language, that Google lobbyists registered in support of chatbot bills in Iowa, Colorado, Nebraska and Arizona according to state lobbying disclosure records, and that identical wording exempting any chatbot that "functions as a speaker and voice command interface or voice-activated virtual assistant for a consumer electronic device" appeared in bills in five states. NPR described the Arizona veto as an instance of "tech-friendly chatbot bills" that did not become law (Source: npr.org). The introduced and House-engrossed versions of HB 2311 contain that exemption word for word; the first Senate amendment changed "functions as" to "incorporates" and added text interfaces (Source: azleg.gov; azleg.gov). The veto letter itself names no company and does not address how the bill was drafted (HB 2311 Veto Letter — Gov. Katie Hobbs (Arizona, June 19, 2026)).

Open questions

• Whether a revised bill is introduced in the 2027 session, and whether it adopts a private right of action or a higher penalty ceiling, is not known.
• NPR's report places Google's lobbying registration in Arizona on "chatbot bills" without identifying HB 2311 by number; the Arizona lobbying disclosure records themselves have not been reviewed.

Relationships

• depends-on: HB 2311 Veto Letter — Gov. Katie Hobbs (Arizona, June 19, 2026) — the Governor's stated reasons for the veto
• depends-on: Arizona HB 2311 — Bill Text (Senate Engrossed with Farnsworth Floor Amendment #2, 2026) — the bill text as passed (Senate Engrossed version with floor amendment #2)
• related: Katie Hobbs — vetoed the bill
• related: Colorado HB 26-1263 (Chatbot Safety Act) — enacted chatbot statute of the same session with a near-identical service definition
• related: Idaho SB 1297 — Conversational AI Safety Act, Nebraska LB525 — Conversational AI Safety Act + Agricultural Data Privacy Act — enacted statutes with the same $1,000 / $500,000 penalty structure and attorney-general-only enforcement
• related: California SB 243 — Companion Chatbots — the California companion-chatbot law the House staff summary cites as background
• instance-of: AI Companions — regulation of services that simulate human conversation and companionship
• related: AI Liability — the veto turned on the bill's remedies and liability limits
• related: Age Verification — subsections H–L regulate age-assurance data and bar conditioning access on digital identification
• related: AI and Children, State-Level AI Regulation
• related: Google DeepMind — NPR reports Google lobbying registrations on chatbot bills in Arizona and other states

Sources

• HB 2311 Veto Letter — Gov. Katie Hobbs (Arizona, June 19, 2026) — the veto letter
• Arizona HB 2311 — Bill Text (Senate Engrossed with Farnsworth Floor Amendment #2, 2026) — bill text: the Senate Engrossed version of June 11, 2026 (azleg.gov) with the Farnsworth Additional COW floor amendment #2 (apps.azleg.gov), together the vetoed text. Introduced and House versions at HB2311P.htm and HB2311H.htm
• (Source: apps.azleg.gov) — bill status: sponsor, committee and floor votes, transmittal, Governor's action
• (Source: azleg.gov) — Senate Research "As Vetoed" fact sheet; also the June 9 fact sheet at S.2311ATT_ASPASSEDADDCOW.DOCX.htm
• (Source: azgovernor.gov) — Governor's legislative action of June 19, 2026
• (Source: npr.org) — NPR investigation of state chatbot bills
• (Source: Raw Sources/Idaho Conversational AI Safety Act (SB 1297).md); (Source: Raw Sources/Nebraska Conversational AI Safety Act (LB525).md) — penalty and enforcement provisions of the Idaho and Nebraska acts, for comparison