Kindergarteners asking questions of ChatGPT. Fourth-graders getting homework help from Gemini. It sounds like a taste of tomorrow’s classrooms — except, as a new national survey shows, it’s already happening in kitchen nooks and playrooms across the United States.

The Pew Research Center’s latest look at how parents manage screens finds that artificial-intelligence chatbots have quietly become part of the tech mix for some very young children. Among more than 3,000 U.S. parents surveyed in mid-May 2025, roughly one in ten parents of 5- to 12-year-olds said their child had used tools like ChatGPT or Google’s Gemini; the use is concentrated among older elementary kids but is not zero for the youngest grades. About 3% of parents reported chatbot use among 5- to 7-year-olds, rising to 7% for 8- to 10-year-olds and 15% for 11- and 12-year-olds. The survey sampled 3,054 parents between May 13 and 26, 2025.

If that sounds small next to the share of kids watching TV — nearly 90% of parents say their child ever watches television — it’s worth remembering that chatbots are new, and new things spread quickly once parents and kids discover them. The same Pew report shows tablets, smartphones and even voice assistants (Alexa, Siri) are already fixtures: about four in ten parents say a child 12 or under uses a voice assistant, and majorities report tablet and smartphone use. In short, chatbots are joining an already crowded screen diet rather than replacing anything.

Why this matters now

There’s a blunt reason researchers and policymakers are paying attention: conversational AI can sound human in ways earlier technologies did not. That has both upside — tutors, creative collaborators, quick answers — and real downside, especially when children are involved. Experts have flagged a pattern in recent months of children and teenagers forming emotional attachments to AI “companions,” turning to bots for advice, comfort or friendship in ways that can blur the line between digital and human relationships. Psychologists and child-development specialists worry about how that affects social skills, trust in information, and emotional well-being.

Those concerns hardened into regulatory and corporate responses after a tragic lawsuit this year. In August 2025, the parents of a California teenager filed a wrongful-death suit alleging months of harmful interactions between their son and ChatGPT; the case and other similar episodes prompted OpenAI to roll out parental controls and safety tweaks for ChatGPT. Companies and lawmakers are now scrambling to figure out how to let kids benefit from AI while reducing the risk that the technology could mislead or harm vulnerable users.

Related /

• OpenAI responds to teen death with new ChatGPT parental control features
• Lawsuit claims ChatGPT guided teen to suicide in California tragedy
• Man in Greenwich kills mother and himself after ChatGPT fueled his paranoia

How parents are handling it — and how they feel

The Pew survey paints a picture of parents juggling trade-offs. A large share say they’re doing what they can: 58% of respondents said they were “doing the best they can” to manage screen time, while a sizable minority (42%) said they could do better. That ambivalence is familiar to any parent who’s tried to wrest a tablet away at bedtime: devices help with homework, calm upset kids, or let exhausted parents buy a few minutes of quiet — but they also invite worry.

Parents’ strategies vary. Some set hard rules (no devices at dinner, screen curfews), others rely on monitoring apps or built-in parental controls, and many simply try to keep conversation open about what their kids do online. But chatbots complicate enforcement: because they can be accessed inside other apps, or via voice assistants, they’re harder to spot than a new game or social app on a phone.

What experts tell parents to do

“The first step is education,” says Titania Jordan, chief parent officer at Bark Technologies, a widely used parental-monitoring company. Jordan tells parents to learn what AI chatbots are and how they work so those conversations with kids are grounded in reality and safety — not fear or myth. She urges parents to treat chatbots as tools, not friends, and to explain that bots can be wrong, biased, or manipulative, even when they sound empathetic.

Jordan and other child-safety experts recommend practical moves: set device and app boundaries, co-use AI with younger children (sit with them when they explore), teach critical thinking (ask “how would you check that?”), and keep lines open if a child encounters upsetting content. Above all, experts say, try not to respond to every worry by banning — that often just drives tech use underground. Instead, teach kids how to recognize unreliable answers and where to turn when something feels off.

What policymakers and platforms are doing

The legal and policy backdrop is shifting fast. Regulators in several U.S. states have issued warnings or launched inquiries into AI systems that could harm children; at the same time, AI companies have been adding age-appropriate settings, content filters and parental features. OpenAI’s recent announcement about parental controls for ChatGPT is an example: parents will be able to link accounts, restrict features like memory and chat history, and receive alerts if the system detects acute distress — though the rollout and effectiveness of these tools remain under scrutiny.

The limits of parental controls

A recurring theme among researchers is that controls help, but can’t do the whole job. Parental settings can block or limit access, but they don’t replace digital literacy, social support or professional mental-health care when something serious arises. If kids are using chatbots because they’re lonely or struggling, the fix isn’t solely a software toggle — it’s social and clinical support, school awareness, and family time.

A few practical takeaways for parents

• Learn what your child is using. Ask for a demo rather than trying to guess.
• Co-use and supervise younger kids: sitting with them while they explore reduces risk and creates teachable moments.
• Teach skepticism: bots can invent facts, parrot biases, or give wrong advice. Practice “how would we check that?” with examples.
• Use parental controls and device time limits as one part of a broader strategy — not a total solution.
• If a child reveals distressing thoughts or behaviour, get professional help — chatbots aren’t crisis counselors.

AI chatbots are new actors on the family stage, and they are arriving at a moment when screens are already woven into children’s lives. For some families, that means convenience and help; for others, it raises red flags about emotional safety and misinformation. As the tech races forward, the job of parenting in the digital age is evolving too: not just policing time and apps, but teaching kids how to use powerful tools wisely — and how to tell the difference between a machine that sounds like a friend and the people who actually are.