AI toys are not safe, say consumer and child advocacy reports : NPR

Child safety nonprofit Fairplay is urging gift givers to avoid buying AI toys for children this holiday season, according to a release advisory released Thursday.

Fairplay, along with other children's and consumer rights groups, say these toys – toys such as plush toys, dolls, action figures and children's robots equipped with chatbots and other artificial intelligence technologies – can be dangerous.

Designed to imitate human behavior and interact with children as if they were friends, the toys offer novelty at a time when artificial intelligence is beginning to permeate many corners of people's lives.

Fairplay's advisory, entitled “AI toys are NOT safe for children,” states that these toys undermine children's trust and harm human relationships, among other things. The recommendations have been endorsed by more than 150 experts and groups, including MIT professor and author Sherry Turkle, pediatrician and researcher Jenny Radesky, the Social Media Victims Law Center and the International Gaming Association of the United States.

“It's ludicrous to expect young children to escape the potential harm here,” Rachel Franz, director of the Fairplay program, said in a statement to NPR. “Young children are particularly susceptible to the potential harms of these toys, such as invading their privacy, collecting data, creating false trust and friendships, and crowding out what they need to thrive, such as human interaction and time to play with all their senses. This could have long-term and short-term development implications.”

Others share concerns

The recommendations follow similar recent warnings from the Public Interest Research Group (PIRG). The non-profit consumer advocacy organization's 40th annual edition of “Trouble in Toyland.” report says some artificially intelligent toys allow for in-depth discussions of sexually explicit topics, have few parental controls and collect reams of data about their underage owners.

“Potentially they all collect your child's votes. They collect his names, his dates of birth. All kinds of information—what a child likes and doesn't like, favorite toys, favorite friends,” Teresa Murray, co-author of the PIRG report and director of the consumer surveillance program, said in the report. interview with NPR. “Because they're connected to the Internet, so everyone is accessible, what could these toys start talking to your kids about with their friends, their friends' parents, or your neighbors? I mean, it's terrible.”

Toy industry and AI players emphasize security, privacy

The toy industry and artificial intelligence companies are responding to such concerns by emphasizing their focus on security and privacy.

OpenAI said it had suspended the maker of the artificial intelligence teddy bear Kumma earlier this week after This is reported by PIRG. the toy shared questionable advice with minors, such as providing details on how to find and light matches. At the request of the researchers, it also discussed sexual issues in detail. “We have blocked this developer [the Singapore-based toymaker FoloToy] for violating our policies,” OpenAI spokesperson Gabi Raila said in an email to NPR. “Our usage policy prohibits any use of our services for the purpose of exploiting, endangering or sexualizing anyone under 18 years of age. These rules apply to every developer who uses our API, and we monitor and enforce them to ensure our services are not used to harm minors.”

The company's technology has also been incorporated into other AI toys, including the AI ​​robotic pet Luna. strategic partnership with Mattel earlier this year “to support AI-powered products and experiences based on Mattel’s brands.” No products have been announced yet. But the company said Mattel's initial set of products and features will be aimed at families and older customers, rather than users under 13.

Fairplay highlighted several AI toys to demonstrate the potential risks of data collection and the impact on a child's understanding of trust. Among them MikoA cute plastic robot with educational games and the slogan “Designed to be your new best friend.” Luna Petbota small plastic companion robot that moves on wheels and has a screen and an ear, and GabboCube-shaped plush robot with big eyes in anime style. It doesn't have a screen, but you can connect it to Wi-Fi and voice chat. NPR has reached out to the manufacturers of these products for comment.

“Children's safety is our top priority,” Curio, the company that makes Gabbo and other artificially intelligent toys, said in a statement to NPR. “Our fences are carefully designed to keep children safe, and we encourage parents to monitor conversations, track information, and choose the controls that work best for their family in the Curio: Interactive Toys app.”

“Facial recognition on Miko 3 is completely optional and exists solely to help families have a more personalized and interactive experience,” Ritwik Sharma, senior vice president of Miko.ai, said in an email to NPR. “It’s important that all visual data is processed locally on the device. They are never sent to the cloud or transmitted externally. To provide families with greater confidence and control, every Miko robot comes with a physical camera shutter that allows parents to completely block the camera whenever they want.”

In an email to NPR, the Toy Association, which represents toy manufacturers, said toys sold by responsible manufacturers and retailers must meet more than 100 strict federal safety standards and tests, including the Children's Online Privacy Protection Act, which regulates children's privacy and data security online and is overseen by the Federal Trade Commission. “The Toy Association encourages parents and guardians to shop only from reputable toy manufacturers, brands and retailers who put children's safety first,” the statement said, adding that it suggests Security Tips for AI and Other Connected Products to better inform people's purchasing decisions.