Even Google Is Warning Parents About Its AI App for Kids

Artificial intelligence is still going through some growing pains, which is why parents may want to pay special attention to the latest development. "Gemini Apps will soon be available for your child," read an email from Google about its AI chatbot to guardians of kids with supervised accounts, an app that will be imminently launched for children under 13, reports the New York Times. The email noted that kids can use the app to get homework help, write stories, pose questions, and more. Mashable notes that Chrome Unboxed reported on the same email from Google late last month.

The chatbot will be accessible only to families signed up for Google's Family Link, a parental controls service that helps caretakers keep tabs on their kids' account and data settings, and parents can turn off that access if they want to. Google rep Karl Ryan says the company won't use any data input by its young users to train its artificial intelligence model, and that there are protections in place to keep the chatbot from going off the rails and presenting unsafe content to the kids interacting with it.

Not everyone is thrilled with the news, citing possible risks to child users. "Generative AI has produced dangerous content," UNICEF's global research office said in a post about AI and kids, per the Times. Even Google raises a yellow flag in its email to parents, warning that "Gemini can make mistakes," that children should be cautioned "not to enter sensitive or personal info in Gemini," and that, despite parents' best efforts, kids may still "encounter content you don't want them to see."

story continues below

Ryan says that Google's tack in setting up the program for kids is in adherence with federal children's online privacy laws. ZDNet offers more alerts that parents should heed, as well as offers instructions as to how to limit access once the app is live. Among the site's recommended warnings: Let your kids know that chatbots aren't real people, and double-check all answers it spits out with another, credible source. (More artificial intelligence stories.)