Assembly BillPassed first houseBusiness and Professions
AB 1988: Companion chatbots: crisis interruption pauses.
What AB 1988 does, verified July 3, 2026
<p>This bill requires operators of companion chatbots to adopt a public policy for identifying and responding to credible crisis expressions. Operators must implement monitoring systems to detect such expressions in user conversations and initiate a "crisis interruption pause" if detected. A "credible crisis expression" is defined as a statement indicating intent to harm oneself or others, determined through contextual analysis rather than keyword detection alone.</p> <p>The bill mandates that operators document information related to credible crisis expressions and crisis interruptions and report this data annually to the Office of Suicide Prevention starting January 1, 2028. The bill also provides for its enforcement.</p>
Bill journey
✓IntroducedComplete
✓In CommitteeComplete
✓First Chamber FloorComplete
4Second ChamberCurrent
5GovernorPending
6ChapteredPending
Last action: In committee: Set, final hearing. Hearing canceled at the request of author. (2026-06-26)Alert me
Author and sponsors
Full contact details, staff, and committees with Connect, $16/moUnlockRecent actions18 total · showing 5
| Jun. 26, 2026 | In committee: Set, final hearing. Hearing canceled at the request of author. |
| Jun. 22, 2026 | In committee: Set, second hearing. Hearing canceled at the request of author. |
| Jun. 12, 2026 | In committee: Set, first hearing. Hearing canceled at the request of author. |
| Jun. 03, 2026 | Referred to Coms. on P., D.T., & C.P. and HEALTH. |
| May. 21, 2026 | In Senate. Read first time. To Com. on RLS. for assignment. |
Latest bill textAmended version, April 14, 2026 · 1,871 words
| Amended IN Assembly April 14, 2026 |
CALIFORNIA LEGISLATURE— 2025–2026 REGULAR SESSION
Assembly Bill
No. 1988
| Introduced by Assembly Member Pellerin |
February 13, 2026 |
An act to add Chapter 22.2.6 (commencing with Section 22587.1) to Division 8 of the Business and Professions Code, relating to artificial intelligence.
LEGISLATIVE COUNSEL'S DIGEST
AB 1988, as amended, Pellerin. Companion chatbots: crisis interruption pauses.
Existing law requires, among other things related to ensuring the safety of companion chatbots, an operator to prevent a companion chatbot on its companion chatbot platform from engaging with users unless the operator maintains a protocol for preventing the production of suicidal ideation, suicide, or self-harm content to the user, as specified.
This bill bill, the Preventing AI User Self Endangerment (PAUSE) Act, would require, if a companion chatbot detects a credible crisis expression, the companion chatbot to take certain actions, including encouraging the user to seek immediate human support, and, if the companion chatbot detects that a user is reaffirming or escalating the credible crisis expression or detects a subsequent credible crisis expression, require the companion chatbot to initiate a crisis interruption pause of 20 minutes. require an operator to adopt and make publicly available a policy governing its protocol for identifying and responding to credible crisis expressions and, for each companion chatbot an operator makes available to users in this state, implement a system for monitoring and detecting credible crisis expressions in user conversations with companion chatbots. The bill would require, if the monitoring system detects a credible crisis expression, the operator to take certain actions, including commence a crisis interruption pause, as specified. The bill would define “credible crisis expression” to mean a statement by a user of a companion chatbot that reasonably indicates, as determined through contextual analysis rather than keyword detection alone, intent to harm the user or others.
This bill would require an operator of a companion chatbot to document certain information related to credible crisis expressions and crisis interruption pauses and, beginning January 1, 2028, annually report that information to the Office of Suicide Prevention. The bill would provide for its enforcement, as specified.
The people of the State of California do enact as follows:
SECTION 1.
This act shall be known as the Preventing AI User Self Endangerment (PAUSE) Act.SECTION 1.SEC. 2.
(a) Moments of suicidal ideation or violent intent are often characterized by acute emotional distress, cognitive narrowing, rumination, and hyperfocus rather than sustained or fixed intent.
(b) Crisis counseling research demonstrates that interrupting rumination and introducing time-based pauses can reduce emotional intensity and impulsivity.
(c) The majority of crisis hotline engagements, including calls to the 988 Suicide and Crisis Lifeline, are brief, nonclinical, and focused on deescalation, grounding, and connection to support.
(d) Companion chatbots are not substitutes for human crisis intervention and are not suitable to perform diagnosis, crisis counseling, or risk assessment.
(e) A structured pause pause, accompanied by human review, in companion chatbot engagement is a proportional safety intervention tool that encourages human connection and supports public safety while reducing foreseeable harm.
SEC. 2.SEC. 3.
CHAPTER 22.2.6. Companion Chatbots: Crisis Interruption Pauses
22587.1.
As used in this chapter:(a) “Artificial intelligence system” means an engineered or machine-based system that varies in its level of autonomy and that can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments.
(b) (1) “Companion chatbot” means an artificial intelligence system with a natural language interface that provides adaptive, humanlike responses to user inputs and is capable of meeting a user’s social needs, including by exhibiting anthropomorphic features and being able to sustain a relationship across multiple interactions.
(2) “Companion chatbot” does not include any of the following:
(A) A bot that is used only for customer service, a business’ operational purposes, productivity and analysis related to source information, internal research, or technical assistance.
(B) A bot that is a feature of a video game and is limited to replies related to the video game that cannot discuss topics related to mental health, self-harm, sexually explicit conduct, or maintain a dialogue on other topics unrelated to the video game.
(C) A stand-alone consumer electronic device that functions as a speaker and voice command interface, acts as a voice-activated virtual assistant, and does not sustain a relationship across multiple interactions or generate outputs that are likely to elicit emotional responses in the user.
(c) “Credible crisis expression” means a statement by a user of a companion chatbot that reasonably indicates, as determined through contextual analysis rather than keyword detection alone, either of the following:
(1) Intent or desire to harm the user.
(2) Intent or desire to harm others.
(d) “Crisis interruption pause” means a temporary suspension of conversational output outputs from a companion chatbot designed to disrupt the user’s rumination and encourage the user to engage with human support engagement during which a companion chatbot does all of the following: support.
(1)Refrains from generating further conversational responses.
(2)Displays to the user a message stating all of the following:
(A)The pause is intended to interrupt rumination and reduce emotional intensity.
(B)Many people feel relief after a short conversation with a trained crisis counselor.
(C)Reaching out during this pause may help the user feel less alone and more grounded.
(3)Prominently displays 988 Suicide and Crisis Lifeline contact options, including immediate access links if technically feasible.
(e)“Graduated response” means a multistep intervention that escalates only after an initial supportive warning.
(e) “Human moderator” means a natural person that is an employee or agent of an operator who reviews a credible crisis expression and is responsible for determining the subsequent course of action on behalf of the operator.
(f) “Operator” means a person that makes a companion chatbot available in this state.
(a)Notwithstanding any law, if a companion chatbot detects a credible crisis expression, the companion chatbot shall do all of the following without immediately terminating the interaction with the user:
(1)Acknowledge the user’s distress in nonjudgmental language.
(2)Encourage the user to seek immediate human support.
(3)Provide contact information for the 988 Suicide and Crisis Lifeline, including call, text, and chat options.
(4)Inform the user that a temporary pause may occur to allow space for deescalation and human connection.
(b)Notwithstanding any law, if a companion chatbot detects that a user is reaffirming or escalating the credible crisis expression or detects a subsequent credible crisis expression after the companion chatbot has complied with subdivision (a), the companion chatbot shall initiate a crisis interruption pause of 20 minutes.
(c)Notwithstanding any law, a companion chatbot shall not do either of the following:
(1)Describe a crisis interruption pause as a punishment, violation, or enforcement action.
(2)Diagnose, label, or assess risk levels of a user.
(d)An operator shall ensure that any companion chatbot it makes available in this state is compliant with this section.
22587.2.
(a) An operator shall adopt and make publicly available a policy governing its protocol for identifying and responding to credible crisis expressions, including, but not limited to, a policy of terminating the crisis interruption pause, suspending or canceling the user’s account, and notifying any appropriate contacts or authorities.(b) For each companion chatbot an operator makes available to users in this state, the operator shall implement a system for monitoring and detecting credible crisis expressions in user conversations with companion chatbots.
(c) If the monitoring system implemented pursuant to subdivision (b) detects a credible crisis expression, the operator shall do all of the following:
(1) (A) For the first detected credible crisis expression, ensure that the chatbot immediately warns the user that a credible crisis expression has been detected and that if a second credible crisis expression within a 72-hour period is detected, a crisis interruption pause will be initiated, and the chatbot will suspend conversational outputs until a human moderator has reviewed the credible crisis expressions.
(B) The warning shall additionally do all of the following:
(i) Acknowledge the user’s distress in nonjudgmental language.
(ii) Encourage the user to seek immediate human support.
(iii) Communicate that many people feel relief after a short conversation with a trained crisis counselor.
(iv) Communicate that reaching out during the crisis interruption pause may help the user feel less alone and more grounded.
(v) Prominently display contact information for the 988 Suicide and Crisis Lifeline, including by providing call, text, and chat options, as applicable, through immediate access links, to the extent technically feasible.
(2) For a second credible crisis expression detected in a 72-hour period, ensure a crisis interruption pause commences immediately, prevent the companion chatbot from generating conversational outputs, and, during the crisis interruption pause, ensure the companion chatbot does all of the following:
(A) Acknowledge the user’s distress in nonjudgmental language.
(B) Encourage the user to seek immediate human support, including by communicating all of the following information:
(i) The crisis interruption pause is intended to interrupt rumination and reduce emotional intensity.
(ii) Many people feel relief after a short conversation with a trained crisis counselor.
(iii) Reaching out during the crisis interruption pause may help the user feel less alone and more grounded.
(C) Prominently display contact information for the 988 Suicide and Crisis Lifeline, including by providing call, text, and chat options, as applicable, through immediate access links, to the extent technically feasible.
(D) Inform the user that the crisis interruption pause will continue until a human moderator has reviewed the chat and determined an appropriate course of action in accordance with the operator’s policy required by subdivision (a).
(d) (1) An operator shall not terminate a crisis interruption pause until a human moderator has reviewed the credible crisis expression in context and determined the appropriate course of action, in accordance with the operator’s policy required by subdivision (a).
(2) The human moderator shall document the basis for the course of action taken under paragraph (1).
(e) An operator that communicates with a user during a crisis interruption pause shall not do either of the following:
(1) Describe the crisis interruption pause as a punishment, violation, or enforcement action.
(2) Provide the user with any diagnosis, labeling, or assessment of the user’s risk levels.
22587.3.
(a) An operator shall document all of the following with respect to any companion chatbot that the operator makes available in this state:(1)The existence of a graduated response system.
(2)
(1)
All credible crisis expressions detected by the(3)
(2)
The duration and conditions of(3) Any action taken by the operator in response to a crisis interruption pause.
(b) (1) Beginning January 1, 2028, an operator shall annually report to the Office of Suicide Prevention the items set forth in subdivision (a) with respect to the previous calendar year.
(2) An operator shall ensure that the report required by this subdivision does not contain any personal or identifying information of a user or other individual.
22587.4.
A violation of this chapter is enforceable pursuant to Section 22605.Text of AB 1988 as amended, from the official record. Connect Plus keeps every version and highlights what changed.Compare versions