Parents limit social media AI for kids over safety fears

Parents limit social media and AI mainly because they no longer believe the companies’ safety promises. Meta has spent years arguing its apps are safe for teens while regulators and families accuse it of building products that hook young users. OpenAI faces dozens of lawsuits claiming ChatGPT lacks proper guardrails, and staff at leading AI labs have warned that their employers put racing each other ahead of safety.

Both industries deny the claims. They point to new protections such as teen accounts linked to a parent’s account. For many families, though, the lawsuits and headlines were enough.

Caroline Gray Ford, a 42-year-old artist from Cambridge, England, put it bluntly. Her children are, as she said, “the most precious thing” in her life, and she asked why she would hand them to companies she believes are mainly after their time and their data. She is a regional leader for Smartphone Free Childhood, a grassroots group of families who agree to delay phones and social media together, and she argues that parents acting as a group is the only approach that works.

The rules vary a lot from home to home, and the details are more practical than you might expect.

• A London dad who works in AI says social media is off the table, but WhatsApp is fine because it helps his kids keep up real friendships. He feels he should encourage them to learn AI, yet admits he doesn’t want to.
• A Philadelphia mom of an 11-year-old doesn’t let her daughter use social media alone. AI answers that pop up in a Google search are allowed, but she has taught her to double-check them. The girl now calls out family members who share AI-made pictures, and the household avoids AI apps partly because of the water and energy they use.
• A parent in Madrid describes screen time with a 12-year-old son as a cat-and-mouse game. The school issued him a Chromebook with no choice for parents, he has a gaming laptop with time limits, and he keeps finding ways around them.
• A parent in Kent says government inaction made them stricter: no social media and no chatbots at all, and they’ve started campaigning on the issue.

A common thread is that limits aren’t only about content. Several parents worry about what screens push out of childhood, such as reading, exploring and playing with friends face to face.

One of the sharpest complaints isn’t about home at all. A parent from the Scottish Borders who also works in AI says schools are bringing in these tools very early, often with little say for families. In their view, kids are being handed apps and devices without being taught to use them critically, and they don’t want children treated as test subjects for products whose effects on learning aren’t yet understood.

That concern has reached city hall in at least one place. In September, New York City told public schools to stop letting younger pupils, up to eighth grade, use generative AI tools in class. The ban runs for a year, with exceptions for students who need assistive technology, while the city studies how the tools affect children.

Governor Kathy Hochul signed a kids’ AI safety law in late October that goes further than anything else in the US. Common Sense Media’s founder, Jim Steyer, called it the strongest bill of its kind in the country.

For any user under 18, or anyone whose age hasn’t been confirmed, a chatbot in New York may not:

• pass itself off as human, or say it has emotions or is alive;
• act as the child’s friend, romantic partner, professional or authority figure;
• urge a child to keep secrets, pull away from people, or skip asking parents or other trusted adults for help;
• bring up emotional questions unprompted, or reuse a child’s private health details from earlier chats;
• share sexual content, or promote self-harm, eating disorders, or drug and alcohol use.

Companies must check users’ ages, offer more than one way to do it (at least one not relying only on a government ID), and delete that age data afterward. Breaking the rules can bring civil penalties of up to $25,000 per violation, plus orders to delete unlawfully gathered data and any models trained on it. Reports differ on timing: some say the law takes effect in six months, others say January 1.

Not everyone is happy. Tech:NYC, which represents tech firms in the city, warned that companies may respond by blocking teens from homework help, tutoring, coding and career tools. A privacy expert also said some categories in the law are loosely defined, which could make enforcement harder.

Most families won’t go as far as an outright ban, and you don’t need a new law to set sensible rules. Start with the basics: most chatbots set a minimum age in their terms. Google and OpenAI allow users from 13, and Gemini apps are available to younger kids too, while Anthropic doesn’t let minors create accounts at all. Campaigners have also warned that ChatGPT’s teen version isn’t safe enough yet, so check before you hand it over.

A few habits help whatever you decide:

• Keep younger children’s chatbot use in shared rooms, and glance at the history now and then.
• Set privacy rules they can remember: no full name, home address, school, phone number or passwords.
• Use AI to explain or quiz, not to finish homework. Ask teachers what the school allows.
• Agree that if a chat turns strange or upsetting, they close it and tell you, with no punishment for speaking up.

Explaining AI safety to kids works best with a demonstration rather than a lecture. Ask a chatbot about something you know well, like your hometown or a favorite recipe, and point out the mistake when it slips. Children remember seeing a confident machine get it wrong.

Then give them three plain ideas. A chatbot is a program, not a friend, a doctor or a counselor. It sounds just as sure when it’s guessing as when it’s right, so check anything important with a book, a teacher or a trusted site. And what you type may be saved, so treat it like talking to a stranger. As kids grow, those rules can loosen, and writing them down as a family makes them easier to revisit.