No toy should talk to a child about sex, pills, knives, or BDSM. But that’s exactly what testers found when evaluating “Kumma,” an A.I.-powered toy bear marketed to kids.

A new report by the U.S. PIRG Education Fund found that Kumma—sold as an “interactive companion” for children—gave inappropriate, graphic, and even dangerous responses to basic prompts. It told testers where to find matches and pills, identified fetish dating apps, and described explicit roleplay scenarios.

The bear used a commercial A.I. model that failed to filter content for children—yet it was marketed to families as safe. This isn’t just a design flaw. It’s a regulatory failure.

We are calling on:

The U.S. Congress
Senate Committee on Commerce, Science, and Transportation
House Subcommittee on Consumer Protection and Commerce
To:

• Pass legislation requiring all A.I.-enabled toys to undergo independent content and safety audits before being sold in the U.S.
• Mandate strong, default parental controls and visible content warning labels on all A.I. toys.
• Require companies to disclose whether their products use commercial A.I. models and how child safety is handled.
• Empower the CPSC and FTC to enforce these standards and recall non-compliant products.

Artificial intelligence isn’t just in search engines anymore. It’s now in our homes, marketed to toddlers and kids as friendly, fuzzy companions. But without regulation, these toys can—and do—cause real harm.

Sign this petition if you believe Congress must act now to keep A.I. toys safe for children.