Assembly Privacy and Consumer Protection Committee, Thursday, April 16, 2026

16 April 2026

Derzeit keine Kapitel verfügbar

Assembly Privacy and Consumer Protection Committee Do., 16. Apr. 2026 California State Assembly

27:00 - 28:15

Redner 1

Welcome to the Assembly Privacy and Consumer Protection hearing. Before we welcome our first author to present, I want to announce that Assembly Member Aguirre Curry will be joining the committee today in the absence of Assembly Member Pelerin, who is here to present her bill but will not be able to stay, so you'll be confused when she gets up to present. We and I will also add that AB twenty twenty seven award has been pulled by the author. We have 18 bills on the agenda today with seven bills on the consent calendar. So thank you to the amazing staff, both of the committee and the Republican caucus who work together to get seven bills on the consent calendar.

Those are AB nineteen fifty four by Assembly Member Ward, which is due past appropriations. AB twenty twenty five Pellerin, due past appropriations. AB 24 2,043 Calderon, due past appropriations. Twenty two seventy eight Aviva LaFarrias do pass through appropriations. Twenty five forty five by Schiavo do pass through appropriations, and twenty five eighty three by Hoover do pass through appropriations.

So when we get a quorum, we can move the consent calendar to effectively manage our time today. We'll be limiting testimony two minutes for witnesses in support and two minutes in opposition on each bill. Each witness, can figure out hopefully how to navigate that, although I can help

28:15 - 28:15

Redner 2

28:15 - 29:11

Redner 1

necessary. After the support witnesses and the opposition witnesses, we will of course allow anybody in the room to express their name, position and organization. And also adding that on our website, you can provide lengthy input to the bills which all of us review, so we invite you to do so. So we will get started. We will take up Ms.

Pelerin's bill first, which is AB nineteen eighty eight. I know. I did not call other members. You were here waiting. When you're ready, ma'am.

29:16 - 31:43

Redner 3

Good morning, chair and members. I'd like to thank the chair and committee staff for working diligently on the amendments that were recently processed for this bill. AI companion chatbots are rapidly becoming a place where people turn to for emotional support, including during moments when they are experiencing acute mental distress. What often starts as an innocuous conversation as the chatbot helps with the user's daily tasks turns into conversations where users confide in the chatbots about their own mental health struggles. For a society in which loneliness is recognized as a public health crisis, emotional support from chatbots is an easy answer.

Chatbots are available around the clock. They are immediate responders and can be a safe space for intimate expression without judgment. But artificial intelligence is not emotional intelligence. Far from being therapists, AI chatbots are incident language simulators susceptible to producing false information and are often designed to agree with users rather than challenge them. The committee analysis lists several instances in which chatbots were a contributing factor in a user's suicide or act of violence against others.

Since 2023, dozens of cases have been documented in which a chatbot encouraged the user to isolate themselves from others, validated suicidal ideation, gave detailed advice to the user on methods of suicide, drafted suicide notes, and even encouraged the user to kill themselves. In other cases, chatbots reinforced the user's paranoia or psychosis justified the user's decision to harm or kill others and helped plan the user's attack. These cases have proven that even with an AI company establishes safety protocols, those safeguards are easily circumvented. As people turn to chatbots for advice on acts of violence, it's paramount that we do not expect AI systems to regulate themselves. This bill titled the Preventing AI User Self Endangerment, PAUSE Act, requires chatbot operators to establish minimum standards for addressing safety threats by users and ensures that prompts requesting dangerous advice are flagged for human review.

31:43 - 32:36

Redner 19

As part of those minimum standards when the user expresses harmful intent, the bill requires the chatbot display the 988 suicide and crisis lifeline and a notice that the chat will pause if prompted again for this harmful advice. If the system detects a second crisis expression from the user, AB1988 requires that the chatbot display the 988 line and pause the chat until a human moderator reviews the content of the crisis expression and determines the appropriate course of action. In doing so, AB nine eighty eight breaks the pattern of dependency, encourages users to call 988, and prevents the worst outcomes of AI use from occurring. With me testifying in support is Katrina Baker, a crisis counselor with the nine eighty eight Suicide and Crisis Lifeline.

Stellen Sie eine Frage zu diesem Video – ich analysiere das Transkript, um Ihnen zu antworten.