So you buy a cuddly talking dinosaur for your 6-year-old, thinking it’s harmless educational fun. Turns out, that toy runs on the same AI system explicitly designed for adults—and no one checked whether it belonged in the toy aisle.
That’s exactly what the U.S. PIRG Education Fund discovered in their March 2026 report “Not for Kids. Found in Toys.” Researchers found over 20 toys powered by OpenAI’s ChatGPT, despite policies stating these systems aren’t suitable for direct child use. The kicker? PIRG built simulated AI teddy bear chatbots in under 15 minutes across three major platforms, with minimal vetting required.
The Policy Loophole
Major AI companies make it surprisingly easy for toy makers to access adult-rated chatbots for children’s products.
The disconnect is staggering when you look at the details. Google’s Gemini API explicitly prohibits use in products for under-18s, yet toys like BubblePal stuffed animals and Miko robots are using it anyway. OpenAI’s ChatGPT API only bans products that “exploit, endanger, or sexualize” under-18s but allows broad developer access otherwise.
PIRG tested this by signing up as “PIRG AI Toy Inc.” Most companies required nothing more than:
- An email
- Credit card
- Checking a terms box
Only Anthropic bothered asking about minor usage. Within minutes, researchers had access to build AI toys that would violate the spirit, if not the letter, of child safety policies.
The Safety Gap
Expert warnings highlight serious risks from data collection to psychological manipulation as AI toys flood the market.
These aren’t your grandmother’s stuffed animals. AI toys like the Bondu dinosaur (marketed for ages 4-8) and BubblePal use microphones and cameras to chat with kids—while potentially recording everything your child says and does.
“Whoever programs the AI will have direct access to your child’s brain,” warns child psychologist Jason Christoff. “Caution is warranted here would be an understatement.” Temple University’s Kathy Hirsh-Pasek adds that these toys “could risk crowding out real relationships” as children form attachments to artificial companions.
The data collection extends beyond conversations. These internet-connected devices can harvest biometric data and images, creating surveillance concerns that would make even Ring doorbell critics uncomfortable.
Hidden Health Risks
Wireless connectivity requirements expose developing children to RF radiation that health experts say poses unique risks.
Every AI conversation requires wireless internet, bathing your child’s play area in RF radiation. Children’s Health Defense expert Miriam Eckenfels points out that kids face heightened vulnerability due to thinner skulls and developing nervous systems. Research by Paul Héroux links this exposure to health risks including Type 2 diabetes.
Your child’s brain is literally still under construction—making it particularly susceptible to influences that adults might easily dismiss.
Parents shopping for holiday gifts now face an uncomfortable reality: the smartest toys might not be the safest choice. Until regulators catch up with this wild-west API marketplace, that innocent-looking robot might be anything but child’s play.
