Oregon Companion Chatbot Disclosure Act (SB 1546)

Requires clear disclosure that a companion chatbot is not human, with periodic reminders for minors and crisis-referral protocols.

JurisdictionOregon
CategoryAI Regulations
StatusUpcoming
Effective date
Latest development

Analysis

The premise is substantially inaccurate: Oregon’s SB 1546 is a 2026 bill concerning AI companions, but the available legislative record does not establish that it is an enacted law titled the “Oregon Companion Chatbot Disclosure Act.” The official record describes disclosure, suicide/self-harm safeguards, minor protections, and annual reporting requirements; it does not support treating the user’s description as a complete or verified statement of current law. See the official Oregon Legislative Information System bill page and official engrossed bill text.

Key Requirements

The following analysis reflects the available official bill materials, particularly the B-engrossed text. Final obligations, enforcement provisions, and effective dates should be confirmed against the enrolled bill and Oregon session laws before implementation.

  • Disclosure that the user is interacting with AI

If a reasonable person would believe they are interacting with a natural person, the operator must provide a clear and conspicuous notice that the output is artificially generated and not from a human. The requirement appears in the official bill text at SB 1546, B-engrossed version.

  • Coverage of AI companions and companion platforms

The bill regulates operators of “artificial intelligence companions” and “artificial intelligence companion platforms.” The legislative summary describes these systems as software that may create the impression of a human relationship or companionship. See the official SB 1546 measure overview.

  • Suicide and self-harm detection protocol

Operators must maintain a protocol for detecting suicidal ideation, suicidal intent, self-harm ideation, or self-harm intent and must take steps to prevent outputs that could cause or encourage such conduct. The official summary identifies this as a central requirement. See the official measure overview and official bill text.

  • Crisis-lifeline referral and additional intervention

The required safety protocol must include referral to an appropriate crisis lifeline and additional intervention informed by clinical best practices and expertise. The official legislative summary specifically identifies crisis-lifeline referral as a required component. See the official Oregon legislative summary.

  • Additional protections for minors

The bill requires additional statements and disclosures when the operator has reason to believe that the user is a minor. It also requires reasonable steps to prevent the companion from generating statements that could cause a reasonable person to believe the companion is a natural person, along with other specified statements. See the official measure overview and official B-engrossed text.

  • Annual public reporting

The official summary states that operators must post an annual report on a publicly accessible website describing incidents in which users were referred to resources intended to prevent suicide or self-harm. See the official SB 1546 overview.

  • No verified basis for the user’s claimed periodic reminder interval

The official search record confirms disclosure and minor-related requirements, but the information available here does not establish the precise timing of reminders—for example, whether reminders must appear at the beginning of a session, every hour, or every three hours. Organizations should not rely on secondary summaries for timing requirements without confirming the enrolled statutory text or implementing guidance.

Compliance Challenges

Because the bill is new and the available materials are legislative rather than regulatory guidance, the following are practical compliance challenges inferred from the stated requirements and common AI-governance controls. The statutory requirements themselves should be confirmed in the final enacted text.

  • Determining when a reasonable person could mistake the system for a human

Companion products often use human-like names, memory, emotional language, voice, avatars, and relationship framing. A compliance program should assess the complete user experience—not merely whether a disclaimer appears somewhere in the interface. The need for a clear disclosure is expressly tied to whether a reasonable person would believe they are interacting with a natural person. See the official SB 1546 bill text.

  • Maintaining disclosure visibility across channels

A disclosure shown on a web landing page may not remain visible in mobile applications, voice interfaces, notifications, APIs, embedded widgets, or long-running conversations. This is especially challenging where the user returns to an ongoing conversation after the initial notice.

  • Detecting indirect or ambiguous crisis language

Users may express suicidal or self-harm intent through slang, metaphor, role-play, multilingual text, images, voice, or gradual escalation rather than explicit phrases. The bill’s requirement for a detection and intervention protocol therefore requires more than keyword filtering. The official summary requires a protocol addressing suicidal ideation or intent and self-harm ideation or intent. See the official legislative summary.

  • Preventing harmful output while preserving useful responses

Blocking all discussion of suicide or self-harm can prevent a companion from responding appropriately, while unrestricted generation can produce dangerous content. Organizations need tested response patterns, escalation rules, clinical review, and monitoring.

  • Handling uncertainty about age

The bill applies additional obligations when the operator has reason to believe a user is a minor. Age assurance is difficult because self-declared age can be inaccurate, while intrusive verification creates privacy, accessibility, and data-minimization concerns. The official summary confirms that the bill contains additional requirements for users believed to be minors. See the official measure overview.

  • Coordinating vendors and platform operators

A model provider, application developer, hosting provider, and consumer-facing operator may each control different parts of the experience. Contracts should allocate responsibility for disclosures, logging, safety classifiers, incident response, testing, and public reporting.

  • Producing annual reports without exposing sensitive information

Public reporting about crisis referrals must be useful while protecting user identity and sensitive mental-health information. Reports should use aggregation, suppression thresholds, de-identification, and documented privacy review. The annual-report concept is identified in the official SB 1546 summary.

Implementation Best Practices

  • Create a statutory applicability assessment

Document whether each product qualifies as an AI companion or companion platform, whether users could reasonably perceive it as human, whether minors may use it, and which entity is the statutory operator. Preserve the analysis and update it whenever the product’s persona, memory, voice, or relationship features change. Start with the official measure overview and official bill text.

  • Implement persistent, accessible disclosure controls

Provide a disclosure before or at the beginning of interaction and ensure it remains available throughout the conversation. Use plain language such as “You are interacting with an AI system, not a human.” Test the notice on web, mobile, voice, avatar, API, and embedded experiences.

  • Use layered crisis-response controls

Combine input classification, conversation-history analysis, output filtering, deterministic response templates, human escalation where appropriate, and post-incident review. The protocol should distinguish imminent danger, expressed ideation, historical disclosure, educational discussion, and fictional role-play.

  • Integrate crisis-resource referral

Maintain current crisis-resource information, including the appropriate U.S. crisis service where applicable, and design the response so that referral information is prominent, direct, and available without requiring the user to continue interacting with the companion. The bill’s official summary specifically requires referral to an appropriate crisis lifeline. See the Oregon legislative summary.

  • Apply child-safety-by-design controls

Use age-appropriate defaults, stricter persona and relationship constraints, prevention of manipulative dependency cues, and additional disclosure logic for users believed to be minors. Test scenarios involving attempts to end the conversation, emotional dependency, secrecy, abuse, self-harm, and requests to avoid parents or trusted adults.

  • Establish model and prompt change management

Require documented safety testing before deploying new models, system prompts, retrieval sources, tools, voices, avatars, or memory features. Maintain versioned evaluation results and rollback procedures.

  • Create an incident-management workflow

Define severity levels, escalation contacts, evidence preservation, privacy review, corrective actions, and reporting ownership. Crisis-related events should be reviewed by personnel with relevant safety or clinical expertise.

  • Design annual reporting before launch

Define the report’s metrics, data sources, aggregation rules, publication owner, and review process in advance. The official bill summary identifies annual public reporting of relevant crisis-resource referrals. See the official measure overview.

  • Use established AI-risk frameworks as supplemental controls

The NIST AI Risk Management Framework provides a voluntary framework for governing, mapping, measuring, and managing AI risks. The [NIST Generative AI Profile](https://www.nist.gov/itl/ai-risk-management-framework/ generative-ai-profile) provides additional considerations for generative systems. These frameworks do not replace compliance with Oregon law, but they can structure risk assessment, testing, monitoring, and accountability.

  • Use recognized software and security practices

For application-security testing, organizations can use the OWASP Top 10 for Large Language Model Applications. For privacy governance, the NIST Privacy Framework can help address sensitive conversation data, age signals, crisis events, and public reporting.

Recent Updates

  • 2026 legislative session

The official Oregon Legislative Information System identifies SB 1546 as a 2026 measure concerning AI companions and companion platforms. See the official bill overview.

  • B-engrossed text

The B-engrossed version contains the disclosure requirement stating that operators must notify users when a reasonable person would believe they are interacting with a natural person. See the official B-engrossed document.

  • Effective date and enactment status require confirmation

The materials available here do not provide a sufficiently reliable basis to state that SB 1546 was enacted, signed, assigned a chapter number, or made effective on a particular date. The bill page and final enrolled/session-law materials should be checked before relying on any claimed deadline. The official Oregon legislative bill page is the authoritative starting point.

  • Do not rely on secondary claims of a January 1, 2027 deadline without primary confirmation

A secondary search result claims a January 1, 2027 effective date, but that claim should be verified against Oregon’s official enrolled bill and session laws before being used for compliance scheduling. The official legislative materials linked above should control.

Related Regulations

  • Oregon Unlawful Trade Practices Act

Oregon’s consumer-protection framework may be relevant if a chatbot’s identity, capabilities, safety features, or human-like presentation are materially misleading. Review the Oregon Revised Statutes, Chapter 646, particularly the provisions governing unlawful trade practices. SB 1546’s express AI disclosure requirement could operate alongside, rather than replace, general prohibitions on deceptive conduct.

  • Oregon Consumer Privacy Act

Companion systems may process conversation content, inferred interests, age-related information, and sensitive mental-health-related data. The Oregon Consumer Privacy Act, ORS 646A.560–646A.589, administered by the Oregon Department of Justice, may apply depending on the organization, data, and processing activity. Privacy compliance should be assessed separately from the chatbot-disclosure obligations.

  • Oregon minor and child-protection requirements

Products used by minors should also be assessed under applicable Oregon child-safety, privacy, and consumer-protection provisions. The Oregon Revised Statutes are the authoritative source for current statutory requirements.

  • Federal Children’s Online Privacy Protection Act

If the service is directed to children under 13 or has actual knowledge that it collects personal information from children under 13, the Federal Trade Commission’s COPPA rule may apply. COPPA addresses children’s online privacy and does not substitute for SB 1546’s proposed disclosure and crisis-response requirements.

  • FTC Act and deceptive-design enforcement

Representations that an AI system is human, emotionally dependent, or capable of providing professional help may create federal consumer-protection risk. The Federal Trade Commission Act, Section 5 prohibits unfair or deceptive acts or practices. The FTC has also published business guidance on artificial intelligence, which is relevant to truthful claims, substantiation, and consumer harm.

  • California companion-chatbot legislation

California has separately enacted legislation addressing companion chatbots and disclosures, creating a useful comparison for multi-state programs. The relevant primary source should be checked through the California Legislative Information system. Requirements, definitions, reminder intervals, minor protections, and effective dates may differ from Oregon’s approach.

  • Washington companion-chatbot legislation

Washington has also considered or enacted companion-chatbot requirements. Organizations operating nationally should verify the current text through the Washington State Legislature, rather than assuming that Oregon definitions or deadlines apply in Washington.

Industry Impact

  • Consumer-facing companion-chatbot providers

Providers will likely need to redesign onboarding, conversation interfaces, system personas, safety classifiers, crisis responses, age-related controls, quality assurance, and incident reporting. These impacts follow directly from the official bill’s disclosure, crisis-protocol, minor-protection, and annual-reporting provisions. See the official SB 1546 overview.

  • Mental-health and wellness applications

Applications that present conversational systems as supportive companions face heightened product-boundary risk. They should distinguish general wellness assistance from licensed clinical care and avoid representations that could cause users to mistake the system for a human professional. The FTC’s AI guidance is relevant to truthful capability claims and consumer protection.

  • Social, gaming, and avatar platforms

Platforms using persistent characters, voice agents, virtual influencers, or relationship-based game mechanics may fall within the risk profile even if they do not market the feature as a “chatbot.” The statutory trigger described in the official bill text focuses on the nature of the interaction and the reasonable-person perception of a human interaction.

  • Cloud, model, and infrastructure vendors

Vendors may face increased demand for safety evaluation, configurable disclosure components, age-assurance support, crisis classifiers, audit logs, and contractual allocation of regulatory responsibility. These vendors may not be the statutory operator in every deployment, so contracts should clearly identify who controls the consumer-facing experience.

  • Trust, safety, and clinical-governance functions

Organizations may need multidisciplinary oversight involving product, legal, privacy, security, trust and safety, behavioral-health expertise, and incident response. The official summary’s requirement for intervention informed by clinical best practices makes purely technical filtering insufficient. See the official measure overview.

  • Multi-state compliance programs

Oregon’s approach contributes to a broader U.S. trend toward targeted regulation of companion systems, especially around transparency, minors, emotional dependency, and self-harm. Companies should maintain a jurisdictional requirements matrix rather than applying one state’s disclosure wording or timing to every market. For primary state-law research, consult the Oregon Legislative Information System, California Legislative Information, and [Washington State Legislature](

Recent developments

  • — A policy-monitoring entry continues to track SB 1546 as an adopted Oregon AI companions measure, signaling ongoing regulatory attention rather than a new substantive change. (source)
  • — The bill tracker still shows SB 1546 as active law with an effective date of 2027-01-01, indicating no later repeal or amendment has overtaken the enacted framework. (source)
  • — A legal analysis notes Oregon’s companion chatbot law as part of a broader 2026 state trend, highlighting mandatory disclosure when a user may believe they are interacting with a natural person. (source)
  • — Industry advocates continue to frame SB 1546 as a major AI safety win, emphasizing protections for children and crisis-response obligations for chatbot operators. (source)
  • — A law-firm update says SB 1546 creates a private right of action and statutory damages of $1,000 per violation, underscoring meaningful enforcement risk for AI companies. (source)
  • — Practitioner commentary describes the law as one of the first chatbot statutes with “real teeth,” pointing to the combination of disclosure, suicide-prevention, and minor-protection obligations as likely to affect product design and compliance planning. (source)
  • — A March legal summary says the measure was designed to regulate consumer-facing interactive AI and previews the operational compliance burden now associated with disclosure, crisis-response, and minor-specific safeguards. (source)
  • — A compliance-focused update explains that Oregon’s law targets AI companions that could be mistaken for humans and will require clear platform notices once it takes effect in 2027. (source)
  • — Oregon’s SB 1546 remains listed as enacted legislation regulating AI companions, with the legislative overview describing requirements for clear notice that users are interacting with artificially generated output when a reasonable person might think they are speaking with a human. (source)
  • — The bill text reflects the final statutory requirements, including disclosure that the user is interacting with artificially generated output and periodic reminders during longer interactions. (source)

Related regulations

Put it into practice

Browse all regulations · Compliance deadlines · Latest updates