Companion chatbots.

Removed the requirement that operators take reasonable steps to prevent companion chatbots from providing rewards at unpredictable intervals or encouraging increased engagement, usage, or response rates (variable-ratio reinforcement). This was a key anti-addiction provision.

Changed the AI disclosure requirement from a blanket obligation to notify users at the start of every interaction and every three hours thereafter, to a conditional requirement that only applies if a reasonable person would be misled into believing they are interacting with a human. This is a substantially narrower trigger.

Changed the suicide protocol language from requiring operators to have a protocol for 'addressing suicidal ideation, suicide, or self-harm expressed by a user' to one for 'preventing the production of suicidal ideation, suicide, or self-harm content to the user.' This shifts the focus from responding to what users say to preventing the chatbot itself from generating harmful content.

Narrowed annual reporting requirements: removed the obligation to report the number of times suicidal ideation was detected by users and the number of times a chatbot brought up suicidal ideation with a user, replacing them with a single metric - the number of crisis service provider referral notifications issued in the preceding calendar year.

Removed the entire requirement for operators to submit their companion chatbot platform to regular audits by an independent third party and to publish a high-level summary of audit results publicly at no cost. This was a major transparency and enforcement mechanism.

The document was converted from the official chaptered PDF format (with page numbers, 'STATE OF CALIFORNIA AUTHENTICATED ELECTRONIC LEGAL MATERIAL' headers, and print pagination) to a web page rendering that includes navigation menus, search fields, version history links, and other website UI elements. The underlying statutory text of Sections 22601 through 22606 is unchanged.

Bill was designated as Chapter 677 of the California statutes upon the Governor's approval on October 13, 2025.

Senate and Assembly passage headers, clerk signatures, and the Governor's transmittal block were removed in favor of the standard chaptered law format.

Standard 'STATE OF CALIFORNIA AUTHENTICATED ELECTRONIC LEGAL MATERIAL' header was added to reflect the bill's final enacted status.

The definition of 'chatbot' was changed from 'user-created character' to 'user-created virtual character' and now explicitly excludes bots used only for customer service purposes, narrowing the bill's applicability.

A new definition of 'artificial intelligence' was added in Section 22601, describing it as an engineered or machine-based system that infers from input how to generate outputs influencing physical or virtual environments.

A new requirement (Section 22602(c)) mandates that operators prevent chatbots from engaging with users unless they have implemented a protocol for addressing suicidal ideation, suicide, or self-harm, including notification referring users to crisis service providers such as suicide hotlines or crisis text lines.

The annual reporting requirements in Section 22603 were restructured, with the language changed from 'all of the following' to 'all both of the following' and items renumbered, suggesting a reduction or reorganization of the specific data points operators must report.

A new private right of action (Section 22606) allows any person who suffers injury in fact due to a violation to sue for injunctive relief, damages equal to the greater of actual damages or $1,000 per violation, and reasonable attorney's fees and costs.

A savings clause (Section 22607) was added stating that the chapter's duties and remedies are cumulative to other laws and do not relieve operators from obligations under any other law.

A standard severability provision was added as Section 2, ensuring that if any part of the act is held invalid, the remaining provisions continue in effect.

The bill's scope expanded from protecting minors to covering all users of companion chatbot platforms. All references to 'minor user' were changed to 'user,' and the chapter title changed from 'Chatbots Used By Minors' to 'Companion Chatbots.'

A new definition for 'companion chatbot' was added: an AI system with a natural language interface that provides adaptive, human-like responses capable of meeting a user's social needs, including by exhibiting anthropomorphic features and sustaining relationships across multiple interactions. The old 'chatbot' definition (which required the character to be 'user-created') was removed.

Operators must now publish details on their suicide/self-harm protocol on their internet website (new subdivision 22602(c)(2)).

The AI disclosure notification requirement was made more specific: operators must remind users the chatbot is artificially generated at the beginning of any interaction and at least every three hours during ongoing interactions, rather than just 'periodically.'

Removed reporting requirements for the number of times a user attempted suicide and the number of times a user died by suicide. The annual report now only requires: (1) number of times suicidal ideation was detected by users, and (2) number of times a companion chatbot brought up suicidal ideation or actions with a user.

The disclosure requirement in section 22605 changed from requiring notice to a minor user and their parent or guardian, to requiring notice to any user that companion chatbots may not be suitable for some minors.

Section 22604 now requires audits to be conducted by an 'independent' third party rather than just any third party, adding a specificity requirement that may affect who qualifies as an auditor.

New Section 22604(b) requires operators to make a high-level summary of audit results publicly available at no cost on their internet website, creating a new transparency obligation that did not exist in the Senate version.

Section 22605 now specifies that the minor-suitability disclosure must appear 'on the application, the browser, or any other format that a user can use to access the companion chatbot platform,' clarifying exactly where the notice must be displayed rather than leaving it unspecified.

Additional coauthors were added: Senator Rubio joined the Senate coauthors, and Assembly Members Lowenthal and Pellerin were added as new coauthors, indicating expanded cross-chamber support for the bill.

Annual reports of suicidal ideation detections must now be submitted to the State Department of Health Care Services Office of Suicide Prevention rather than to the department generally.

A new requirement was added obligating the Office of Suicide Prevention to post data from companion chatbot annual reports on its internet website, creating a public transparency mechanism.

A new definition for 'Office' was added, referring to the Office of Suicide Prevention established pursuant to Section 131300 of the Health and Safety Code.

The standalone definition of 'chatbot' was removed. The customer service exclusion that previously applied to 'chatbot' was folded into the 'companion chatbot' definition as a new subsection (2).

Senator Stern was added as a coauthor, indicating broader legislative support for the bill.