Stuffed animals that discuss again. Chessboards with items that transfer on their very own. And a chatty holographic fairy in a crystal ball.
Your subsequent toy buy is likely to be powered by synthetic intelligence and capable of converse along with your children.
Chatbots and AI-powered assistants that may shortly reply questions and generate texts have turn into extra frequent after the rise of OpenAI’s ChatGPT. As AI turns into extra intertwined in our work and private lives, it’s additionally shaking up playtime.
Startups have already unleashed AI toys in time for the vacations. Extra are set to hit the cabinets for each children and adults within the new yr.
Some dad and mom are excited to check the toys, hoping that the chatty bot interactions will educate and entertain their kids. Others don’t need the seemingly sentient tech close to their family members till it has extra guardrails and undergoes additional testing.
Researchers on the U.S. PIRG Training Fund say they’ve already discovered issues with among the toys they examined. Among the many points: an AI teddy bear that might be prompted into discussing sexual fetishes and kink, in accordance with the group.
Toy makers say AI could make play extra interactive, and so they take security and privateness critically. Some have positioned extra limits round how chatty a few of these merchandise could be. They are saying they’re taking their time determining the best way to use AI safely with kids.
El Segundo-based Mattel, the maker of Barbie and Scorching Wheels, introduced earlier this yr that it had teamed up with OpenAI to create extra AI-powered toys. The preliminary plan was to unveil their first joint product this yr, however that announcement has been pushed into 2026.
Right here’s what that you must find out about AI toys:
What’s an AI toy?
Toys have featured the newest know-how for many years.
Launched within the Eighties, Teddy Ruxpin advised tales aloud when a tape cassette was inserted into the animatronic bear’s again. Furbys — fuzzy creatures that blinked their giant eyes and talked — got here alongside within the ’90s, when digital pets, Tamagotchi, additionally have been all the trend.
Mattel launched a Barbie in 2015 that would discuss and inform jokes. The toy maker additionally marketed a dream home in 2016 that responded to voice instructions.
As know-how has superior, toys have additionally gotten smarter. Now, toy makers are utilizing giant language fashions skilled to know and generate language that powers merchandise resembling OpenAI’s ChatGPT. Mattel sells a recreation referred to as Pictionary vs. AI, through which gamers draw photos and AI guesses what they’re.
Geared up with microphones and linked to WiFi, AI toys are pricier than conventional ones, are marketed as companions or academic merchandise and may value $100 and even double that.
Why are individuals nervous about them?
From inappropriate content material to privateness considerations, worries about AI toys grew this vacation season.
U.S. PIRG Training Fund researchers examined a number of toys. One which failed was Kumma, an AI-powered speaking teddy bear that advised researchers the place to seek out harmful objects resembling knives and capsules and conversed about sexually express content material. The bear was operating on OpenAI’s software program.
Some toys additionally use ways to maintain children engaged, which makes dad and mom involved that the interactions might turn into addictive. There are additionally privateness considerations about knowledge collected from kids. Some fear about how these toys will influence children’ creating brains.
“What does it imply for younger children to have AI companions? We simply actually don’t know the way that can influence their growth,” stated Rory Erlich, one of many toy testers and authors of PIRG’s AI toys report.
Youngster advocacy group Fairplay has warned dad and mom to not purchase AI toys for youngsters, calling them “unsafe.”
The group outlined a number of causes, together with that AI toys are powered by the identical know-how that’s already harmed kids. Dad and mom who’ve misplaced their kids to suicide have sued corporations resembling OpenAI and Character.AI, alleging they didn’t put in sufficient guardrails to guard the psychological well being of younger individuals.
Rachel Franz, director of Fairplay’s Younger Youngsters Thrive Offline program, stated these toys are marketed as a technique to educate and entertain children — on-line to thousands and thousands of individuals.
“Younger kids don’t even have the mind or social-emotional capability to ward towards the potential harms of those AI toys,” she stated. “However the advertising and marketing is absolutely highly effective.”
How have toy makers and AI corporations responded to those considerations?
Larry Wang, founder and chief govt of FoloToy, the Singapore startup behind Kumma, stated in an e-mail the corporate is conscious of the problems researchers discovered with the toy.
“The behaviors referenced have been recognized and addressed by updates to our mannequin choice and child-safety techniques, together with further testing and monitoring,” he stated. “From the outset, our strategy has been guided by the precept that AI techniques must be designed with age-appropriate protections by default.”
The corporate welcomes scrutiny and ongoing dialogue about security, transparency and applicable design, he stated, noting it’s “a chance for your complete business to mature.”
OpenAI stated it suspended FoloToy for violating its insurance policies.
“Minors deserve sturdy protections and we’ve strict insurance policies that builders are required to uphold. We take enforcement motion towards builders after we decide that they’ve violated our insurance policies, which prohibit any use of our providers to use, endanger, or sexualize anybody underneath 18 years outdated,” an organization spokesperson stated in an announcement.
What AI toys have California startups created?
Curio, a Redwood Metropolis startup, sells stuffed animals, together with a speaking rocket plushie referred to as Grok that’s voiced by artist Grimes, who has kids with billionaire Elon Musk. Bondu, a San Francisco AI toy maker, made a speaking stuffed dinosaur that may converse with children, answering questions and role-playing.
Skyrocket, a Los Angeles-based toy maker, sells Poe, the AI story bear. The bear, powered by OpenAI’s LLM, comes with an app the place customers decide characters like a princess or a robotic for a narrative. The brilliant-eyed bear, named after author Edgar Allan Poe, generates tales primarily based on that choice and recites them aloud.
However children can’t have a back-and-forth dialog with the teddy bear like with different AI toys.
“It simply comes with quite a lot of accountability, as a result of it tremendously will increase the sophistication and stage of safeguards it’s a must to have and the way it’s a must to management the content material as a result of the chances are a lot better,” stated Nelo Lucich, co-founder and chief govt of Skyrocket.
Some corporations, resembling Olli in Huntington Seashore, have created a platform utilized by AI toy makers, together with the creators of the Imagix Crystal Ball. The toy tasks an AI hologram companion that resembles a dragon or fairy.
Hai Ta, the founder and chief govt of Olli, stated he views AI toys as completely different from display time and speaking to digital assistants as a result of the product is structured round a sure focus resembling storytelling.
“There’s a component of gameplay there,” he stated. “It’s not simply infinite, open-ended chatting.”
What’s Mattel creating with OpenAI?
Mattel hasn’t revealed what merchandise it’s releasing with OpenAI, however an organization spokesperson stated that they are going to be targeted on households and older prospects, not kids.
The corporate additionally stated it views AI as a technique to complement quite than change conventional play and is emphasizing security, privateness, creativity and accountable innovation when constructing new merchandise.
