California’s New Rules for AI Chatbots and Access to Human Support

California’s new customer service chatbot law will require certain large businesses to make their automated support systems more transparent—and make it easier for customers to reach a human.

Assembly Bill 1609 was signed into law by Governor Gavin Newsom on September 28, 2026.  Scheduled to take effect January 1, 2027, AB 1609 adds a new chapter, entitled “Customer Service Chatbots,” to the California Business and Professions Code. Although its title focuses on customer-service chatbots, its human-access requirements extend beyond chatbot interactions.

For covered businesses, the central compliance challenge is access: ensuring that automation does not become an obstacle to meaningful human assistance. The law permits AI-powered customer service, but establishes rules governing chatbot identification, escalation requests, response times, and customer-service contact information.

Which Businesses Does AB 1609 Cover?

AB 1609 applies to a “large private business,” defined as a business with more than $500 million in gross annual revenue nationally that provides goods and services to customers. The threshold is national revenue—not California revenue—and businesses with exactly $500 million do not meet the statute’s “more than” requirement. A “customer” is a natural person residing in California, with residence determined by the customer’s account address.

The law is not confined to technology companies or businesses headquartered in California. Depending on their revenue, customer base, offerings, and applicable exemptions, covered businesses could include large retailers, subscription providers, consumer-facing software companies, and other enterprises serving California consumers. The statute expressly contemplates customer-service operations conducted through websites, applications, and telephone systems. Its definitions also encompass service and technical assistance for prospective and existing customers.

An important limitation concerns free and freemium services. The statute defines “services” as services provided for monetary consideration, including subscriptions, fees, and recurring charges. It excludes services—or parts of services—that offer a free option with an opportunity to pay for the service or an upgrade later. Businesses offering a mixture of paid and free services should therefore examine the statutory definition rather than assume that every customer interaction falls within the law.

The phrase “private business” should not be treated as an express exemption for publicly traded companies. The operative definition turns on revenue and the provision of goods and services, not on whether a company’s shares trade publicly.

What Does AB 1609 Require?

AI Chatbots Cannot Masquerade as Humans:  AB 1609 establishes two related but distinct requirements. First, a covered business may not represent that a customer-service chatbot is human. Second, when a reasonable person interacting with the chatbot would likely be misled into believing that the interaction is with a human, the business must clearly and conspicuously disclose that the chatbot is artificially generated and not human. The disclosure must appear in the same medium as the interaction.

The law defines a customer-service chatbot as an AI system with a natural-language interface that provides adaptive, human-like responses and performs customer-service or support functions. This definition matters: the law does not simply classify every automated menu or scripted interface as a chatbot. Businesses must evaluate the system’s functionality, not merely its product name.

The statute does not prescribe a mandatory disclosure script. Nevertheless, businesses should review chatbot introductions, agent names, avatars, and conversational language together to assess whether the overall presentation misleadingly suggests a human representative. A disclosure addresses the affirmative notice requirement; it should not be paired with contradictory representations that the chatbot is human.

Customers Must Have a Simple Route to Human Assistance: During at least their regular business hours, covered businesses must provide a simple feature that allows customers to request a human customer-service agent within 15 minutes or offer a specific appointment within one business day. Once connected, individual holds should not exceed 15 minutes, and cumulative waits should not exceed one hour, subject to the law’s good-faith standard..

The obligation is broader than adding a “speak to an agent” button to a chatbot.  The human assistance requirement applies to every platform through which a business may offer customer service, including telephone menus and online support channels. Outsourcing support is permissible, but the statutory obligation remains directed at the covered business. Requests for human assistance made by email, web-based contact form, or voicemail are expressly excluded from the 15-minute connection or response requirement.

Enforcement and Penalties

A California prosecutor may enforce AB 1609. Violations carry civil penalties of up to $5,000 for an initial violation and up to $10,000 for each subsequent violation. These are statutory maximums, not automatic penalty amounts. The statute does not expressly establish a per-customer, per-conversation, or per-day method for counting violations.

AB 1609 expressly creates no private right of action, meaning California residents do not receive a standalone right to sue under the new law.  The law does, however, preserve duties, remedies, and obligations under other laws, so the absence of a new private right of action does not eliminate independently available legal theories for those plaintiffs seeking to sue violators in court.

AB 1609 vs. California’s Existing Bot Disclosure Law

California’s existing bot-disclosure law, SB 1001, became operative on July 1, 2019. Codified at Business and Professions Code Sections 17940–17943, it prohibits using an online bot to communicate with a person in California online with the intent to mislead that person about the bot’s artificial identity, for the purpose of knowingly deceiving the person about the communication’s content to incentivize a commercial purchase or sale or influence a vote in an election.

To avoid liability, companies employing automated bots for the purposes covered by the law must make a clear and conspicuous disclosure reasonably designed to inform recipients that they are communicating with a bot.

AB 1609 builds on California’s existing bot-disclosure law but differs in two important respects: it imposes an objective disclosure standard for covered customer-service chatbots, rather than requiring proof of deceptive intent, and it adds requirements for access to human customer support. It supplements rather than replaces the existing law.

Likely Impact on Covered Businesses

AB 1609 is likely to affect customer-service architecture as much as chatbot disclosures. Covered businesses should assess whether customers can find and use human-escalation features, whether callback systems offer specific appointment times, and whether support platforms can track waits across transfers and escalation queues. These operational questions follow directly from the statute’s access and timing requirements.

The good-faith standard also makes documented processes important. Staffing plans, queue-monitoring procedures, vendor service commitments, callback records, and outage documentation can help demonstrate that a business adopted commercially reasonable and practicable standards intended to achieve compliance. The law does not expressly mandate those particular records, but they can help substantiate the statutory defense.

The likely result is not the abandonment of automated customer service, but a shift toward systems that combine transparent automation with dependable human escalation. Businesses that have relied on automation to limit human contact may face staffing, technology, and vendor-management changes. Those with accessible support channels and effective callback systems may have a more manageable implementation path before January 1, 2027.