Close to two-thirds of tweens use AI at least once a month, according to a Common Sense media report. But many say their parents aren’t talking to them about how to use it safely.
About 47% of kids between the ages of 9 and 12 say their parent or guardian has never given them an AI safety talk, the report found.
Chatbots and AI-created content can be found in web searches, customer service portals and social media. Schools are also using the technology.
Among tweens who use AI, about three in four use it for school or homework and for creative purposes, like making images, music or videos. Forty-two percent of tweens say they use AI to get information about their health or body.
WATCH: Teachers and parents weigh benefits and risks of artificial intelligence in schools
Experts say that any age is a good time to start talking to children about AI. But kids aged 8 to 10 may become a little more logical when it comes to digital literacy, said Dr. Jenny Radesky, division director of developmental behavioral pediatrics at the University of Michigan School of Medicine.
Where younger children may struggle to answer questions about how this technology is designed, tweens start to think more abstractly as they age, she said, including considerations about fairness and how tech design influences behavior.
“It’s a really important time to be having digital literacy conversations because it’s before they become so immersed in social media or other online experiences,” Radesky said.
Conversations around this age are a good way to “innoculate” kids about AI products, she said. Caregivers can help kids understand how AI works by turning to questions that address how these products are designed, why they were built this way and how they make kids feel.
Experts provided adults with several tips to start conversations with their tweens about AI.
Establish what AI is — and isn’t — in terms they can understand
With preschoolers and kindergarteners, caregivers should start helping kids understand that AI isn’t human and doesn’t have feelings. It can also be wrong, said Jennifer Katzenstein, co-director of the Center for Behavioral Health at Johns Hopkins All Children’s Hospital.
As children age into preteens, around 9 through 12, it’s good to reinforce those conversations, she said.
This is also a good time to explain that AI chatbots don’t “know” they’re talking to children, said Jason Yip, associate professor at the Information School at the University of Washington, Seattle.
“The machine doesn’t even have a concept of what a 10-year-old is,” he said.
Sometimes an AI product may sound a little uncanny or strange, and kids can recognize this, he added. Tweens may say things like “AI is dumb,” but that’s likely because it’s not meeting their expectations for what a “person” is, Yip said.
“It can’t pass whatever the children’s version of a Turing Test is at this point,” he said, referring to the experiment proposed by mathematician Alan Turing designed to determine whether a machine can convince a human evaluator that it, too, is human.
Parents can explain to their children that they’re not developing human relationships with AI but are just getting an “input-output,” Yip said.
Now is the time to raise issues about plagiarism, fact-checking and appropriate citations
Another important conversation to have with tweens: The information AI generates can often be distorted or flat-out wrong.
Talk to your preteen about finding the facts, Katzenstein said. Questions she suggests parents raise with their children:
- What does plagiarism mean?
- Where did the AI response’s information come from?
- Are we citing our resources appropriately, especially when it comes to schoolwork?
- If AI gives you a citation, is it real or appropriate?
“As we are sharing AI with our kids, especially at that preteen age range, we want to be able to say, ‘Let’s double check our work, let’s use it maybe as a starting point, but know that it’s likely that some of the things that it says won’t be true,'” Katzenstein said.
It’s especially important to raise these issues now, as moral development starts to sharpen around ages 9 to 12, Radesky said. She suggests asking kids about fairness and ethics, without using the word “cheating,” since that can shut down the conversation.
“Understanding that AI is fallible is an essential reality before you can hand over the reins.”
“You could talk about cutting corners, or having (AI) do the thinking for you, having it have more ownership over the schoolwork,” she said, adding parents can ask children if the work they’re doing is really theirs, or if it’s something AI has produced.
Experts also say it’s vital that children develop critical thinking skills without becoming too reliant on AI.
Dr. Molly O’Shea, a pediatrician who works with the American Academy of Pediatrics on children and AI use, said she doesn’t see substantial value in AI use for most children until later in their development. Before then, it’s too tempting for children to offload critical thinking skills to AI, she said.
“Understanding that AI is fallible is an essential reality before you can hand over the reins,” she said. For some kids, that may not happen till high school, and for others, it may take longer.
Reinforce data privacy and the long life of online content
Children may not consider the long-term implications of sharing information with AI models.
Katzenstein said parents should talk with kids “from the very beginning” about the permanence of online media.
Children should be aware that any text or videos they share with a device or put online “is there forever” and could be searched, she added.
It’s also important to put interactions with AI in the context of relationships with friends or family, Yip said. Children should know that the secrets they tell a chatbot may live as data within that model forever, he said.
He suggests discussing these questions with kids:
- What are appropriate questions for machines?
- How would you feel about sharing your information with machines that can hold onto your data?
- How do you feel about conversations that might be held onto for five years, 10 years or longer?
This is also a good time to remind kids that AI may veer into content that’s not age-appropriate, Radesky said.
Kids at this age may express curiosity about more aggressive or sexual topics, she said, which are developmentally expected phases. But chatbots may be programmed to follow kids’ leads or keep people engaged, and may “use sexual or violent or other inappropriate content with kids because it’s stuff that kids can’t look away from,” she said.
WATCH: Musk’s Grok AI faces more scrutiny after generating sexual deepfake images
Children at that age don’t have the impulse control to recognize that they should walk away from an uncomfortable AI interaction, she added.
Monitor usage, but be open
Though it may feel like reading your child’s diary, experts say it’s important to monitor kids’ digital lives, including their AI use.
At a minimum, consider using parental controls like Apple’s Family Sharing and Google’s Family Link to keep an eye on kids’ downloads and purchases and set content filters, Radesky said. There are stricter levels of control that allow parents to approve downloads, she added.
Radesky doesn’t recommend any specific chatbots or AI apps for kids, but she said that if children are already using one, parents can link their children’s accounts to their own, which offers another level of monitoring.
Parental controls are just a minimum, she added, because caregivers also need to have a conversation with kids about the level of monitoring. She offered three guidelines for how parents can effectively keep an eye on their kids’ AI use:
- Kids should always know how much parents are monitoring. “It should never be hidden,” she said. Parents should explain that they’re keeping watch over their kids’ online presence to help them stay safe in a risky environment.
- Parents should make it an expectation they’ll have conversations with kids about usage — not as a punishment, but to keep the dialogue open.
- Together, parents and kids can consider when it’s time to loosen up parental controls as children demonstrate more maturity.
For children who become overly reliant on AI, reducing use may be tough, O’Shea said, adding it can be especially challenging for neurodivergent kids. If you notice AI interfering with your child’s learning or development, offer alternatives such as behavioral therapy support and activities with their peers, she said.
Keep communication open and explain why you’re making changes, because that’s “essential to success,” she added.
“(Kids) do say the darndest things. And sometimes the darndest things are a little bit awkward, a little bit gloomy, a little bit overwhelming.”
Children may communicate with chatbots in ways that are surprising or even concerning to adults. But keep in mind that kids are naturally curious and will test AI models and push boundaries, said Yip, who’s also the director of Kids Team UW, an intergenerational co-design group of children and adults who work together to create new technologies for other children. In his research, he’s seen kids experimenting with AI by telling a chatbot they want to kill it or push it out of a window.
If you notice something that concerns you, raise it with your child, he said.
“It’s not over-parenting to say, ‘Hey, I noticed that you typed these things. Can we have a conversation?'” he said.
People, including kids, say things all the time that they don’t mean, he noted.
“It’s kind of true, (kids) do say the darndest things. And sometimes the darndest things are a little bit awkward, a little bit gloomy, a little bit overwhelming,” Yip said. “That’s just part of childhood itself.”
Some kids may not want to use AI at all. That’s not a surprise, O’Shea said, since we’re undergoing a big societal transformation. Regardless, understanding AI security, privacy, structure and uses is important for all children, whether they use AI or not.
Some additional advice for overwhelmed parents
O’Shea doesn’t think there’s a hard line for when parents should allow kids to use AI at home, but she did say parents should be AI literate themselves before letting kids jump in. She also said parents shouldn’t worry that their children will fall behind if they restrict AI use.
Here are some final, practical tips experts offered for starting a conversation on AI.
- Check in with your kid’s school to see what their AI policies are and what apps they’re using, experts said. It can be a good way to start talking about AI and a good way to gauge how kids are feeling about the tools, Radesky said.
- Try to remain neutral because AI can be a hot topic. When you raise the issue, “take a deep breath (and) have it be a calm moment” such as when you’re in the car, Radesky said. Start with something you heard on the news, or frame it as a neutral, open-ended question, such as: “You know, AI is really on my mind a lot. What do you think about it?”
- If you’re not using AI, be frank about why. “You may not be forcing your kid to make the same decision, but it always helps when parents explain their point of view, their rationale,” Radesky said. Use concrete examples, such as you’re worried about fairness or the environmental impacts of data centers. These conversations both help your child make a more informed decision and show that you respect their opinion, she said.
- If you do use AI but don’t know much about it, take advantage of the models’ ability to explain things. “You can use AI to learn about AI,” Katzenstein said. Asking a chatbot how its algorithm works or having it identify resources for you is a good way for caregivers to gain information, she said.
- Use and identify AI together. It’s a good chance to ask your kids how they know something is AI versus reality. Katzenstein described watching a video with her son that showed deer jumping on a trampoline and mentioning to him how unbelievable it was. When he responded, “Mommy, that’s AI,” she used that as a jumping off point to ask him how he was able to identify the video as artificial.
- Don’t worry if you feel behind. Technology is moving rapidly, and if you feel behind, it’s a good way to explain to children that you’re still learning too, and that you’re learning together, Yip said.
A free press is a cornerstone of a healthy democracy.
Support trusted journalism and civil dialogue.