Meta Platforms Inc.‘s and OpenAI Inc.‘s personal AI agents, featuring cartoonish mascots, are raising concerns that they’ll appeal to children and put them at risk. The default for Meta’s Muse AI assistant is a furry character named Jolly, while OpenAI’s Dots are characterized by soft, abstract, and colorful shapes.Privacy and children’s safety advocates are sounding the alarm that the AI assistants’ design could make children see them as toys. “It’s one thing for companies to anthropomorphize their products — that is, I think, a manipulative but pretty common thing to do — but they didn’t have to make it so cute, cuddly, and have these very child-appealing features to it,” said Calli Schroeder, senior counsel at the Electronic Privacy Information Center. The products’ releases come as AI makers already are grappling with litigation over chatbot harm to mental health, she said.Advocacy groups and lawmakers have also warned that AI toys could disrupt socio-emotional development by encouraging unhealthy attachments. “The way they’ve designed these agentic AI systems kind of brings the worst of all worlds in AI development into one spot,” Schroeder said. Cartoon DesignFor their part, the companies say the agents are designed to help adults. Muse has already racked up more than 7 million downloads since its debut in early September, Meta said. The Facebook parent also unveiled a small gadget, called the Muse Charm, that allows users to access the AI assistant while away from Wi-Fi.Meta said in a blog post that “it felt very odd talking to a corporate logo or entity,” and Meta employees began creating their own avatars. “We require everyone using the Muse app to provide their date of birth, we block people we detect may be under 18 from creating an account, and we do additional checks to help ensure people who say they’re adults are actually over 18,” a Meta spokesperson said. “The agent is designed to make users’ lives easier by booking flights, scheduling appointments, and making reservations — and it’s clear people are loving it.”Dots aren’t currently available to users under 18, and the default representation is a neutral gray circle, according to information provided to Bloomberg Law by OpenAI.Yet there’s early criticism that the agents’ appearance belie the assurances that they’re meant for adults.They “look a lot like a character you would see on any given kid’s morning cartoon,” said Haley Hinkle, policy counsel at Fairplay, a children’s advocacy nonprofit. “It really does feel like it’s supposed to further these companies’ goals of making chatting with a robot as if it is a trusted friend and confidant more comfortable for adults and children.”Both companies have faced backlash for fostering unsafe online environments for children. A New Mexico jury in March ordered Meta to pay $375 million for misleading teenagers about the the safety of its platforms, while a California jury found Meta partially liable in a landmark social media addiction trial the same month. And OpenAI is facing mounting litigation over its alleged role in fostering mental health crises in teens. Schroeder said “it’s really surprising” considering that backdrop that personal AI assistants meant for adults “seem so intentionally appealing to children.”Not ‘Wild West’The companies aren’t lacking in experience protecting children’s safety. “I think sometimes this gets portrayed as the Wild West, and it’s not necessarily,” said Jennifer Huddleston, a senior fellow in technology policy at the Cato Institute. “There is at least some existing framework when it comes to children’s data” through the 1998 Children’s Online Privacy Protection Act, which regulates the online collection of personal information from children under 13, she said. The corresponding US Federal Trade Commission rule says children’s images and voices are covered by the law, so companies that make AI products that listen and process kids’ speech must comply with it. COPPA established multiple factors that determine whether an online service is aimed at children: subject matter, use of animated characters, and whether advertising is directed toward them. Products like AI toys or chatbots directly aimed at kids between eight and 13 might implicate COPPA, but Muse and Dots are general-use products, Huddleston said. The notion that COPPA is triggered because the AI agents are designed to look like toys is “taking it a little far,” she said. COPPA’s definition of what’s directed to children means “a lot of times we’re looking for characters that are popular with children, or things that resemble that,” like a Barbie doll with a built-in webcam, she added.But Meta isn’t going into product design with a blank slate, Schroeder said. “We would have to be embarrassingly naive to believe that it just didn’t occur to Meta that this design is appealing to children,” she said. “I don’t think it’s out of line to say that a billion-dollar company should have some people on staff that know better, particularly when they have been repeatedly called out for manipulative design to children.”State RegulationRegardless, states are beginning to regulate AI products explicitly aimed at kids, Hinkle said.California enacted a suite of youth online safety measures this year, including a ban on toys with built-in chatbots for kids under 16. The New York legislature has passed a similar moratorium, and the Pennsylvania House is considering an AI toy ban as well. Those laws focus on physical toys, so it’s not yet clear whether they cover Meta’s and OpenAI’s personal assistants, Hinkle said. Her organization last year advised against AI toys, saying they could expose children to sexually explicit or dangerous conversations. But “every child and every family is unique; they have unique values and they have unique ideas about what age is appropriate for certain access to technology,” Huddleston said. “Our goal should be to educate and empower parents and young people to be able to make the right decisions for themselves and their families around technology, not to have the government dictate what technology can and can’t be allowed.”
————————————————
Source link
