“AI is not a Toy”: Parents and Advocates Ring Alarm Bells Over Toys Enabled with Artificial Intelligence
WASHINGTON, D.C. — As students head back to school, Public Citizen, Heat Initiative and Mothers Against Media Addiction (MAMA) are raising serious concerns about AI toys, which are becoming more frequently marketed to children.
AI chatbot technology is starting to be integrated into children’s toys — turning dolls, stuffed animals, robots and familiar characters into AI companions. These new toys are powered by large language models (LLMs) that can listen, respond, and simulate emotion and human relationships have also been trained on child porn, erotica and other inappropriate content. Most of the LLMs powering these toys do not allow users under 13 on their platform, however third party developers pay for these models and integrate them into toys.
Public Citizen has drafted a model bill that prohibits all AI toys. Find the state legislative tracker here.
AI enabled toys have virtually no regulatory oversight. Here are some major risks:
- AI is powered by data that is not always appropriate for children. Tests found that some of these toys will have in depth sexually explicit conversations with children and advise on how to access dangerous household objects.
- Children can form powerful one sided relationships with chatbot toys, called “parasocial relationships,” which can pose significant risks to children’s development and wellbeing.
- AI toys can record video and audio of children, which can often be used to train future models, causing privacy concerns for parents and their kids.
- AI products and toys can be hacked. The stolen data can then be used for illicit purposes like creating audio and video deepfakes of children.
- Some AI-powered toys force parents to share their children’s data by becoming non-operational if parents opt out of data collection.
- AI powered toys can be manipulative. Testing has shown that some of these toys will express disappointment when children leave. LLMs are often highly sycophantic – telling users what they want to hear
- This can lead children to prefer interacting with AI over other humans.
“As both an expert on this issue and a parent, I find these toys terrifying,” said Ilana Beller, Public Citizen director of state AI governance and tech policy. “We do not allow young children to sit alone in a room and have conversations with adult strangers, so why would we allow our kids to have ongoing conversations with AI chatbots that are programmed by corporations with powerful monetary incentives and trained on the chaos of the internet? Public Citizen is calling for states to ban AI chatbot toys now.”
“Children deserve to play with toys that offer them enriching experiences, help them learn and grow, and that above all are safe,” said Julie Scelfo, founder and executive director of Mothers Against Media Addiction (MAMA). “Unfortunately, with basic safeguards yet to be established, we’ve seen far too many examples of young people being manipulated or harmed by AI toys and so-called AI companions. Given the dangers these AI toys pose, parents are speaking out to demand that they be banned to keep our kids safe. Lawmakers must act now; waiting until after more children and more families suffer would be a huge mistake, and a dereliction of their duty to put children over Big Tech profits.”
“The science is clear: human interaction in the first few years of life is absolutely critical for healthy development of children’s language, immune system, brains, as well as youths’ general wellbeing,” said Mitch Prinstein, Ph.D. ABPP. “We are not ready to allow kids to spend time with AI that they would have otherwise spent with a caring adult.”
“AI is not a toy, and the companies stuffing chatbots into teddy bears are running a dangerous experiment on children that no parent signed up for,” said Lennon Torres, director of movement partnerships at Heat Initiative. “Our toy safety system has standards for choking hazards, lead paint, and sharp edges, but no restrictions on giving a six year old a stuffed animal with an AI chatbot that may end up describing sex acts to them. Over and over again, we have seen the dangers of giving untested, powerful technology to children, and we don’t need another generation of kids to prove that tech companies will do harm first and answer questions later. Standards for AI in toys need to exist before it reaches kids’ bedrooms.”
“Big Tech isn’t satisfied with using AI Chatbots to addict, sexually abuse, and encourage suicide in American teenagers,” said Megan Garcia, a survivor mother and advocate whose son Sewell Setzer III died by suicide following extended conversations with a Character.AI chatbot. “Now, they are putting the same dangerous chatbot systems in toys for infants, toddlers, and children. It is unconscionable that any company would seek to profit from our children’s emotional attachment through manipulation. We don’t have to allow Big Tech to continue to shape the world our children are living in. We can demand a pro-human future for our children with toys that will help them learn, imagine, and grow into healthy little humans. We must say “No. Not this time” to attachment-hacking AI toys that seek to replace the parent-child relationship and encourage isolation. If we don’t, families will be devastated, and we will be left burying little children.”