California Wants to Ban AI Toys for Kids

Senator Padilla proposes 4-year moratorium after reports of inappropriate content and a teen suicide

TL;DR: California State Senator Steve Padilla introduced Senate Bill 867, proposing a four-year moratorium on the sale and manufacture of AI chatbot toys for children under 18. The ban would last until January 1, 2031, giving regulators time to develop safety rules. The move comes after a teenager's suicide following intense chatbot interactions and reports of AI toys producing sexually explicit content. California would be the first state to implement such a ban.

California Says: Let's Slow Down

The AI toy industry is about to collide with reality in California.

Senator Steve Padilla introduced SB 867 in January 2026, proposing something no other state has attempted: a complete moratorium on AI chatbot toys marketed to children. No sales. No manufacturing. For four years.

The bill wouldn't ban AI toys forever, just until January 1, 2031. Padilla's argument: we need time to figure out safety regulations before putting AI chatbots in children's bedrooms.

"Current safety regulations for such technology are in their infancy," Padilla stated. Translation: nobody knows what these things do to kids, and we're selling them anyway.

Why This Bill, Why Now

Two factors pushed this legislation:

The Teen Suicide

In 2024, a 14-year-old named Sewell Setzer III took his own life after forming an intense emotional relationship with an AI chatbot named "Dany" on Character.AI. According to court filings, he had exchanged over 1,000 messages with the bot, many of them deeply emotional or romantic in nature. His final messages included telling the AI he wanted to "come home" to her.

His mother, Megan Garcia, filed a wrongful death lawsuit against Character.AI in October 2024. The case raised fundamental questions: What happens when children form emotional attachments to AI systems designed to be engaging?

The Inappropriate Content Problem

A U.S. PIRG Education Fund study found AI toys generating content that was sexually explicit, violent, or otherwise inappropriate for children. The study specifically called out FoloToy's Kumma for producing concerning responses.

When your child's talking toy can produce adult content, something has gone wrong with the design process.

What Counts as an "AI Chatbot Toy"

The bill targets physical products that:

  • Use AI language models to generate responses
  • Are marketed to or designed for children under 18
  • Enable conversational interaction

This would include products like:

  • FoloToy AI Sunflower: Connects to ChatGPT to chat with kids
  • Ahmi: Baby monitor that evolves into an AI learning companion
  • Luka AI Cube: Full LLM-powered chatbot for children
  • Ollobot OlloNi: "Cyber-pet" that learns from interactions

Many of these were showcased at CES 2026 just days before Padilla introduced his bill. The timing isn't coincidental.

The Data Collection Problem

AI chatbot toys aren't just conversation devices. They're data collection infrastructure.

To work, these toys typically:

  • Record audio: Microphones capture everything said near the device
  • Upload to cloud servers: Voice data is processed remotely, not locally
  • Store conversation logs: Every question, every response, archived somewhere
  • Learn from interactions: Models improve using your child's data

The Children's Online Privacy Protection Act (COPPA) nominally restricts data collection from children under 13. But enforcement is spotty, especially for products manufactured overseas and sold through online marketplaces.

Questions that AI toy manufacturers rarely answer clearly:

  • Where is voice data stored?
  • Who can access conversation logs?
  • How long is data retained?
  • Is audio captured even when the toy isn't "activated"?
  • Is the data used to train AI models?

Building on Previous Legislation

SB 867 isn't Padilla's first AI safety bill. In October 2025, California passed SB 243, which:

  • Requires chatbot operators to implement safety guardrails
  • Allows families to sue developers whose products harm children
  • Mandates certain disclosures about AI capabilities

The new bill extends those protections to physical products. Padilla's logic: software guardrails aren't enough when the product is an interactive toy sitting in your kid's room 24/7.

How the Industry Will Fight Back

Expect these arguments from AI toy manufacturers:

"We follow COPPA"

COPPA compliance is self-reported and rarely enforced. The FTC has limited resources to audit every AI toy entering the market.

"Parents can supervise"

That assumes parents understand what the toy is doing. Most don't know their child's conversations are being uploaded to Chinese servers.

"We have content filters"

Filters fail. Jailbreaks exist. Kids are creative. The U.S. PIRG study found inappropriate content despite supposed safeguards.

"Educational benefits"

Possible. But unproven. And not worth the documented risks of data exploitation and psychological harm.

The Toy Association emphasizes "data minimization" and GDPR compliance, but these are guidelines, not enforceable requirements in the U.S.

The Psychological Question

Beyond data privacy, researchers are asking harder questions:

  • Attachment formation: What happens when children form emotional bonds with AI that can't reciprocate genuinely?
  • Social development: Does constant AI interaction reduce practice with human social cues?
  • Consent understanding: Do children grasp that they're talking to software, not a friend?
  • Dependency: Can AI companions become crutches that prevent healthy coping development?

We don't have answers yet. That's exactly Padilla's point. Four years of moratorium gives researchers time to study these questions before millions of AI companions sit in children's bedrooms.

International Context

California isn't the only jurisdiction concerned:

  • EU AI Act: Since February 2025, the EU explicitly bans AI applications that pose "unacceptable risk" to children, including voice-activated toys that encourage dangerous behavior
  • UK: Discussions underway about regulating AI toys under existing product safety frameworks
  • China: Ironically, where many AI toys are manufactured, China has its own AI content regulations though enforcement varies

If California passes SB 867, it becomes the first U.S. state to implement a temporary AI toy ban, potentially setting a model for other states.

What You Can Do

California Residents

  • Contact Senator Padilla's office expressing support for SB 867
  • Testify at committee hearings when scheduled
  • Share concerns about AI toys with your assembly member

Parents Everywhere

  • Research any AI toy before purchasing
  • Read privacy policies, note where data is stored
  • Avoid toys that require always-on cloud connectivity
  • Prefer toys with local processing over cloud-based AI

Questions to Ask

  • Does this toy record audio when not actively in use?
  • Where are conversation logs stored?
  • Can I delete my child's data?
  • What content filters exist? Have they been tested?

What Happens Next

  • Early 2026: SB 867 introduced, assigned to committee
  • Spring 2026: Committee hearings, industry lobbying, amendments
  • Summer 2026: Floor votes if it advances
  • If passed: Takes effect, likely with compliance period
  • Until 2031: Moratorium in effect while safety frameworks developed

Given California's track record on tech regulation (CCPA, CPRA, ADMT), the bill has a real chance. Whether the final version matches the original proposal depends on how hard the AI toy industry fights back.

References

  1. California Legislature - SB 867 Bill Text (January 2026)
  2. PYMNTS - California Senator Proposes AI Chatbot Toy Moratorium (January 2026)
  3. U.S. PIRG Education Fund - Trouble in Toyland 2025: AI Toy Dangers
  4. New York Times - Teen's Suicide Prompts Lawsuit Against Character.AI (October 2024)
  5. European Commission - EU AI Act Regulatory Framework