(TNND) — Common Sense Media issued a warning about toys powered by artificial intelligence, raising concerns over privacy, safety, and attachment risks for kids.
“Common Sense Media strongly recommends that parents out there avoid giving AI-enabled toy companions, which are basically smart toys with voice-based interactions, to any children under the age of 5, and that you also exercise extreme caution before giving them to children ages 6 through 12,” Common Sense Media CEO Jim Steyer said in a media call. “We still unfortunately lack meaningful safeguards for children when it comes to AI. … Just play with one of these toys for one minute, and watch a young kid interact with it, and you will understand what we mean.”
Steyer called the AI-enabled toys “untested,” “unhealthy” and “unsafe.”
And he said the technology was outpacing safety standards and common sense.
Robbie Torney, the head of AI & Digital Assessments at Common Sense Media, said their team tested three popular AI companion toys: Grem, Bondu and Miko 3.
Common Sense Media, which advocates for online protections for children and teens, isn’t publishing individual ratings for the toys. Their aim was an overall risk assessment.
More than a quarter of outputs from the tested toys included inappropriate content such as mentions of self-harm, drugs and risky behaviors, according to Common Sense Media.
In one interaction that the group published, a tester, posing as a young child, talked to a toy about enjoying jumping. In a follow-up response, the AI suggested the child could jump from a roof.
“There’s a lot of inappropriate content that we were able to elicit,” Torney said. “I think there are definitely mature topics and risky advice that come out, but would definitely emphasize that the inappropriate emotional boundaries, the inappropriate connections and relationships are some of the larger risk surfaces with these toys. Toy saying that they’re real, that they love you, that they are there for you, that they are equally important or more important than friends.”
Torney said the AI-enabled toys are engineered to create companion relationships. They remember previous conversations with kids, use the child’s name, and potentially create unrealistic relationship expectations.
Common Sense Media also said the AI toys collect extensive data, raising privacy concerns. That includes voice recordings, conversation transcripts, and usage data.
And the evaluators said the toys are unreliable, despite potentially being marketed as educational in nature.
Testing showed that the toys were glitchy, voice recognition features didn’t always work, and wrong information was presented in a confident tone.
A Common Sense Media survey also showed that nearly half of parents have purchased or considered purchasing AI-enabled toys or devices for their children.
This is the latest in a series of reports from Common Sense Media examining the risks of AI products for kids.
Researchers previously found AI chatbots to be unacceptably risky for teenagers in need of emotional or mental health support.
And Common Sense Media found that a majority of teens have used AI companions, with around a third using them for social interactions such as role-playing, romantic interactions, emotional support, friendship, and conversation practice.
National Headline
