Blog

AI Chatbot Disclosure Laws 2027: What SMEs Must Show

New Colorado and California rules take effect on January 1, 2027. Here is which business chatbots and voice agents must say they are AI, and how a small business builds that in.

Atiq Rezaul Hoque Turjo
Atiq Rezaul Hoque Turjo
8 min read
AI Chatbot Disclosure Laws 2027: What SMEs Must Show

Your chatbot should not make customers guess whether they are talking to a person. But a new law does not automatically mean every small-business bot is covered.

The ai chatbot disclosure law picture changes on January 1, 2027, when new rules begin in Colorado and California. Colorado excludes many commerce and customer-service tools; California's new customer-service requirements target very large businesses. An older California rule can still affect businesses of any size when bots are used deceptively to drive sales.

What changes on January 1, 2027?

Three rules matter, but they do different jobs.

Colorado HB26-1263: Approved May 29, 2026, and effective January 1, 2027, this Colorado conversational AI law sets disclosure and safety duties for operators of covered public-facing conversational AI services. For covered services, the AI disclosure must appear at the beginning of the user's first interaction each day, at least every three hours during a continuous interaction or remain persistently visible, and whenever the user asks whether the service is AI. Additional duties address self-harm, professional claims and known minor users. Annual reporting on the crisis protocol starts July 1, 2027. Read the Colorado General Assembly bill summary and signed statutory text.

California AB 1609: Approved September 28, 2026, this law takes effect January 1, 2027. It regulates customer-service chatbots operated by a defined large private business, with a revenue threshold far above a typical SME. It prohibits presenting those bots as human and requires disclosure and a route to human assistance in specified circumstances. See the California AB 1609 bill text.

California Business and Professions Code 17941: Already operative since July 1, 2019, this separate California bot disclosure law addresses bots used online with intent to mislead people about their artificial identity to incentivize a sale or influence a vote. A clear, conspicuous disclosure provides a defense. It is not limited to large businesses. Read Section 17941.

Does my chatbot count under Colorado HB26-1263?

Start with the bot's primary purpose, not whether it uses an LLM. Colorado defines a conversational AI service broadly as a publicly accessible system primarily designed to simulate human conversation and interaction, then excludes several categories.

A restaurant booking assistant, salon appointment bot or online store order helper is likely excluded when it is primarily built for commerce or transactional assistance. That exclusion expressly includes recommendations, shopping, ordering, payments, delivery, returns and customer support. A typical AI customer-service agent may fit this category, but its actual behavior matters.

An employee-only HR policy assistant or internal SOP search tool is likely excluded because the law excludes tools used solely internally and certain business operations, productivity, research, training and technical-assistance tools. A clinic's AI receptionist operated for a HIPAA-covered entity may fall under the specific HIPAA exclusion; check the operator and deployment rather than assuming every health-related bot qualifies.

A public, open-ended companion that invites visitors to discuss anything is more likely covered. A lead-generation bot that begins with service questions but freely becomes a general-purpose companion is harder to classify. The statute also excludes certain narrow, single-topic tools with restrictions on sexual content and self-harm dialogue. If your assistant's scope crosses categories, check with counsel before relying on an exclusion.

Bot or agent Colorado HB26-1263 (from 1 Jan 2027) California AB 1609 (from 1 Jan 2027) Good practice anyway
Website order or booking bot Likely excluded as transactional assistance Only if a large private business and within customer-service scope Identify AI before taking details
Customer-support chat Likely excluded as customer service Only if a large private business Keep an AI label and escalation option
AI receptionist / voice agent for a clinic that is a HIPAA covered entity Likely excluded under HIPAA provision; check with counsel Only if a large private business and within scope Spoken disclosure and staff transfer
Internal staff assistant Likely excluded as internal use Likely outside customer-service scope Tell employees what tool they use
Open-ended public “ask anything” chat May be covered; check with counsel Only if a large private business and within scope First-message and persistent disclosure
Lead-gen chat that wanders beyond the business's services May be covered; check with counsel Only if a large private business and within scope Constrain topics and offer human handoff

The table is a screening tool, not a scope determination. Review what the system actually does, who operates it, and whom it serves.

Ukiyo-e print of a small shop entrance with an indigo curtain and a hanging bell at daybreak.

A clear introduction at the entrance sets expectations before the conversation begins.

Who must comply with California AB 1609?

AB 1609 is not a blanket mandate for every California small-business chatbot. Its definition of a large private business uses a national gross annual revenue threshold far above a typical SME. The new requirements concern covered customer-service chatbot interactions.

A covered business must not represent its customer-service chatbot as a human. When a reasonable person is likely to be misled, the business must clearly and conspicuously disclose that the customer is interacting with AI in the same medium. For a voice interaction, that means a spoken disclosure rather than relying only on a website banner.

The business must provide a simple way to request a human agent during regular business hours. It must make a good-faith effort to connect the customer within 15 minutes or offer an appointment within one business day, subject to the law's hold-time requirements and exceptions. Enforcement is assigned to public prosecutors; the law does not create a private right of action.

For SMEs, the immediate lesson is about product expectations. A franchise, software supplier or platform serving large enterprises may add AI labels, handoff controls and queue handling that smaller customers can adopt too. Meanwhile, the existing California Section 17941 rule still matters to any business using a bot with the specified intent to mislead people into a sale. Small does not mean exempt from that older rule.

Want to check the bot already on your website or phone line? Book a free chatbot disclosure review with NextCTL. We can map the introduction, labels and human-handoff path before you change the workflow.

How should we design an AI disclosure that works across both states?

Build AI disclosure requirements into the experience rather than hiding them in a privacy policy. A few simple patterns make the bot understandable to customers and easier for your team to audit.

Start with a direct first message. For example: “I'm an AI assistant for [business]. I can help with bookings and common questions. You can ask to speak to our team.” Put it before the assistant collects a detailed request, not after a long exchange.

Keep the identity visible. A chat header reading “AI assistant” is useful even after the welcome message scrolls away. For a Colorado-covered service, a persistent visible disclosure can satisfy the recurring-disclosure alternative during a continuous interaction; otherwise the statutory timing requirement still needs attention. Reset the first-interaction logic appropriately for each day.

Say it aloud on voice calls. An AI voice agent can open with: “Hello, I'm the AI receptionist for [business].” A caller should not have to infer the system's identity from an unnatural pause or synthesized voice.

Answer identity questions directly. If someone asks, “Are you a person?” the answer should be “No, I'm an AI assistant,” not a joke or evasive response. This is particularly important for Colorado-covered systems, where responding to such prompts is an express requirement.

Make human help discoverable. A human handoff chatbot flow should be easy to find: add “Talk to a person” in the chat interface and a voice command such as “representative.” Define what happens when the office is closed, no agent is available or the customer requests a callback. For businesses subject to AB 1609, test the statutory connection and appointment rules rather than treating a button alone as compliance.

In AI chatbot development, these should be acceptance criteria: first-message disclosure, persistent label, identity-question test, handoff trigger, unavailable-agent fallback and logs showing which path ran.

Ukiyo-e print of an incense timekeeping box beside an open back-room door with an attendant waiting.

A lasting reminder and an open route to staff help people stay oriented throughout the exchange.

What about sensitive topics, crisis messages and professional claims?

Colorado's covered conversational AI operators have obligations beyond a simple “I'm AI” line. They must maintain a suicide and self-harm protocol that includes referral to crisis services. They also must not represent AI outputs as coming from, or being equivalent to those of, a licensed health-care, legal or mental-health professional or a qualified dietitian. Additional obligations apply when the operator knows a user is a minor.

For a business assistant, design a boundary: it can explain office hours or help arrange an appointment, but should not impersonate a doctor, lawyer or therapist. When a message suggests immediate danger, trigger the appropriate crisis-referral response rather than continuing a sales script. When a customer needs professional judgment, route the matter to qualified staff.

Even if your bot is likely excluded from this particular Colorado law, these controls reduce confusion. An AI governance review should inspect the actual prompts, escalation paths, knowledge sources and fallback messages—not just the disclaimer on the homepage.

How do these disclosure rules relate to other AI laws?

Colorado HB26-1263 is a different Colorado law from the Colorado AI Act, which addresses a separate set of high-risk AI system obligations. California's automated decision-making rules for AI agents are also a different regime; telling someone a chatbot is AI does not, by itself, resolve obligations related to automated decisions.

Treat disclosure as one control in a broader inventory of what your AI does, what decisions it affects and where people can intervene.

What should be on our 2027 bot disclosure checklist?

Use this for each public-facing website chatbot, AI receptionist and voice assistant—not just for the software platform as a whole.

  • [ ] List every public-facing and internal conversational AI deployment and its operator.
  • [ ] Document each bot's primary purpose, audience and whether it is commerce, customer service or open-ended chat.
  • [ ] Screen Colorado exclusions and California AB 1609 business-size and customer-service scope; flag uncertain cases for counsel.
  • [ ] Check whether California Section 17941's misleading-identity and sales-intent rule could apply.
  • [ ] Add a plain AI introduction at the start of chat and a spoken introduction for voice.
  • [ ] Keep a persistent chat identity label or implement required repeat disclosures where applicable.
  • [ ] Test direct answers to “Are you AI?” and “Are you a real person?”
  • [ ] Provide a visible human-handoff option, with clear after-hours and unavailable-agent handling.
  • [ ] Test crisis referrals, professional-claim boundaries and known-minor safeguards where relevant.
  • [ ] Record test results, assign an owner and revisit the workflow whenever its purpose or capabilities change.

These are practical design checks, not a substitute for a legal assessment of your deployment. This article provides general information, not legal advice.

If you're launching a chatbot or reviewing one already in production, book a free consultation with NextCTL LLC. We can help make disclosure, routing and human handoff part of the system design rather than a last-minute banner.

FAQ

Frequently Asked Questions

FAQ

Do small businesses have to disclose AI chatbots in 2027?

Not every small-business chatbot is covered by the new 2027 rules. Colorado excludes several common commerce and customer-service uses, while California AB 1609 applies to businesses above a large-business revenue threshold. However, California's older bot law can still apply regardless of size when a bot is used with intent to mislead about its identity to incentivize a sale. Clear identification remains a sensible default.

Does Colorado's chatbot law apply to customer service bots?

Colorado HB26-1263 expressly excludes software primarily designed for commerce-related or transactional assistance, including customer support and customer service. A conventional order-status or appointment-support bot is therefore likely excluded. But a public bot that regularly moves into open-ended conversation may require closer review. Classification depends on its primary design and actual operation, so uncertain deployments should be checked with counsel.

Who does California AB 1609 apply to?

California AB 1609 applies to defined large private businesses, using a national gross annual revenue threshold far above that of a typical SME. For covered customer-service chatbots, it requires honest AI identification, a same-medium disclosure when users are likely to be misled, and a simple way to request human assistance during regular business hours. Its connection, appointment and hold-time provisions also matter.

Does an AI voice agent have to say it is AI?

For a covered California AB 1609 interaction, disclosure must be in the same medium when a reasonable person is likely to be misled, so a spoken line matters. Colorado-covered conversational AI services also have specific disclosure timing and identity-question requirements. Whether a particular receptionist is covered depends on its purpose and operator. Regardless, opening with “I'm the AI receptionist” avoids confusion and makes handoff easier.

Continue Reading

Independent check

Decide with evidence, not a pitch.

Open a neutral research prompt in the AI you already use. It asks where NextCTL fits, where a freelancer or an in house hire may be enough, and what to verify before you choose.

AI answers can be incomplete. Check nextctl.com/methodology before deciding.

See where your operations
are leaking revenue.

Tell us what you want to fix. We show where missed calls and slow follow up cost you most, with a written plan for what to fix first.

Or run a free tool first:Website AuditAI Readiness Scan