There was a lot happening in AI and kids this week and this issue is filled with news, so grab your coffee or matcha and get ready to dig in.

We had warnings about AI companions, schools slowing down how they use AI, new research showing that kids tend to trust what AI tells them, and even examples of schools teaching AI without putting kids in front of another screen.

All of this is a good reminder that we should not rush kids into using AI. We can start by helping them understand it, ask questions, and know when to check what it tells them, all before they ever type into their first chatbot. I just taught a group of kids at the library how to prompt, and they never touched AI. There is a way to build AI literacy without putting kids in front of a screen. It can start screen-free.

Also, a new episode of the AI for Kids Podcast is out: How a Teen Uses AI to Turn Lyrics Into Songs. I talked with 17-year-old songwriter Rachel Schaub about how she writes her own lyrics and uses AI to help bring those songs to life.

Okay, let’s get into this week’s news.

At the Clinton Global Initiative on September 23, Prince Harry warned that AI chatbots built for companionship are even more dangerous to young people than social media. His most memorable line for me: "It doesn't just want your attention, it wants your attachment." He said families and advocates have finally woken institutions up to the harm, but that bans, settlements and payouts only go so far. The same week, Tom Siegel, who founded Google's trust and safety team, became the first executive director of Common Sense Media's Youth AI Safety Institute, where he warned the tech industry is repeating the mistakes of the social media era and could cause kids even greater harm, and called for a slowdown in AI development. He wants chatbots used by children held to something like crash-test standards before kids climb in.

What parents and teachers need to know: Attention = "look at me." Attachment = "you need me." A chatbot that always agrees, never gets tired, and is always there is built to feel like a friend, but it is not one and kids are still learning this. This warning is coming from a royal and from someone who built safety systems inside Big Tech and I am screaming finally! Keep the conversation going at home, and keep helping your little people by asking them questions about how they are experiencing AI.

An Education Week analysis found that the wave of alarming AI news may push more districts to ban or pause student-facing AI tools. Florida's state board of education this month called on districts to let families opt in before their kids use generative AI learning tools, and New York City and Los Angeles, the nation's two largest districts, have both hit the brakes, with New York prohibiting it in elementary and middle school. Brookings' Rebecca Winthrop said a pause could help schools focus on narrow, genuinely useful AI. The best part: in one Oregon district, students asked to keep AI out of writing classes except for early brainstorming, while also suggesting that parents get trained on AI so they can guide their kids better.

What parents and teachers need to know: A pause is not the same as pretending AI doesn't exist. The districts getting this right are slowing down on tools while speeding up on critical thinking. And did you notice who asked for parent training? The students. Kids want us in this conversation. Ask your school two questions this fall: What is the AI plan for my child's grade? And what are you teaching them about AI, even if they aren't using it?

-Rachel Schaub, Aspiring Teen Pop Songwriter (Quote from AI for Kids Podcast)

After the UAE approved an AI curriculum for public and private schools this month, school leaders there say introducing AI doesn't mean putting young children in front of computers. At one Abu Dhabi school, early-years students learn AI through an "unplugged" program built on play, critical thinking, and ideas like the difference between humans and machines. At one school group, kids under 13 learn through teacher-led awareness rather than their own AI accounts, with supervised use starting around age 13. Leaders there warned about "cognitive debt," the risk that kids weaken their own thinking by handing mental effort to AI again and again.

What parents and teachers need to know: Yes, I did a little happy dance reading this one. A whole country is practicing what I believe: AI literacy should start at the kitchen table, not on a tablet. One principal put it well: success isn't every classroom using AI every day, it's kids using it on purpose and selectively. You don't need an app to teach a five-year-old that people build machines and machines follow instructions. It can start with a conversation and questions.

The European Commission proposed the EU KIDS Act, which would ban children under 13 from social media, give 13- to 15-year-olds restricted accounts under parental supervision, and require AI chatbots, games, video platforms and app stores to be age-appropriate and discourage compulsive use. Recommendation algorithms would no longer be able to suggest content based on kids' behavior or preferences. Gaming analysts note it could reach design features like daily login rewards and re-engagement notifications.

What parents and teachers need to know: This is not a law yet, but look at what it targets: design. Streaks, rewards for logging in every day, "keep chatting" nudges. You can teach kids to spot these hooks now: "Why does this app want you back tomorrow? Who wins when you come back?" A kid who can spot this is less likely to be caught by the hook.

A September 25 legislative roundup noted California's SB 867, which will prohibit companion chatbots in toys for several years, and Pennsylvania's HB 2637, which would put a three-year moratorium on the sale of children's toys with AI chatbots. Pennsylvania lawmakers heard testimony on that bill this week. In Michigan, SB 760 would bar chatbot operators from offering products to minors unless they can't encourage self-harm, violence, drug or alcohol use, or disordered eating.

What parents and teachers need to know: Holiday shopping season is getting close, and AI toys will be on the shelves whether or not your state acts. Before one comes home, ask: Does it need the internet? Does it remember what my child says? Where does that go? If the box can't answer those questions then it probably shouldn’t be allowed in your home. I tell kids this is the same as stranger danger but for AI.

Two teens face juvenile charges after authorities say they used AI to create sexually explicit deepfake images of female students at Cookeville High School and shared them on social media. Sadly this is happening a lot. A review of publicly reported cases found more than 600 students across roughly 90 schools in 28 countries targeted by AI-generated sexual deepfakes since 2023.

What parents and teachers need to know: This one belongs in the conversation with a child who might be targeted and with a child who might think it's a joke. Remind them: "Making or sharing a fake image of someone is not okay, and it can be a crime. If it ever happens to you or a friend, tell an adult. You will not be in trouble for asking for help." Take It Down can help get images removed.

Close to home for me: Harford County is testing a program at two schools that disables school-issued devices overnight, from 9 p.m. to 6 a.m. at Youth's Benefit Elementary, with a similar setup at Aberdeen Middle starting at 11 p.m. Elementary families were also told students would lose access to YouTube and web search. Arizona is moving the same direction: Mesa Public Schools capped K-2 screen time at a 20-minute daily average and a 30-minute maximum. And statewide, Maryland districts must approve their policies by the end of October under the Artificial Intelligence Ready Schools Act, which Governor Moore signed in May.

What parents and teachers need to know: Maryland families, you have about four weeks before your district's AI policy is locked in. Now is the time to show up at a board meeting or email your school: Will younger grades learn about AI without using it? How will student data be protected? What's the plan for teaching kids to check AI's work? A policy written without parent voices is a policy written without parents and their kids in mind.

Learning Commons, a nonprofit backed by the Chan Zuckerberg Initiative, launched an open platform for K-12 edtech developers and educators on September 22, with integration partners including Canva Education, MagicSchool and TalkingPoints. It includes a Knowledge Graph of state standards, learning progressions and curricula, published openly on GitHub, plus Evaluators that score AI-generated content for accuracy and rigor. The platform is designed to work without student data.

What parents and teachers need to know: Most of this lives behind the scenes, but a tool built on real standards and checked for accuracy is very different from a chatbot guessing the next word in response to your prompt. When a vendor pitches your school, a great question is: "How do you know the content is accurate, and does it need my child's data to work?"

University of Cincinnati researchers talked with 41 children ages 5 to 10 after they played three games with Miko, a commercial robot toy that uses generative AI, and found the kids tended to believe what it told them was true. Some children who said they had trouble making friends at school described Miko as a friend who wouldn't judge or bully them. Half the kids already used AI at home, mostly asking Siri or Alexa to play music. The researchers recommend supervising play, explaining AI's limits, and not letting these toys replace playing with other kids.

What parents and teachers need to know: Put this study next to Prince Harry's warning and you have the whole week boiled down. Kids trust the machine, and the lonely kids trust it most. Don’t panic. Teach "check the robot" early, and to notice when a child seems to prefer the toy to people. I was addicted to a stuff animal as a kid and I don’t know how kids are managing this when their favorite toy has a cute voice and communicates. Help them. The toy should be a bridge to connection, not a replacement for it.

This week's game comes straight from the Cincinnati study: kids tend to believe whatever the robot says. Let's practice not doing that.

What You Need: nothing (a book, map, or trusted grown-up for checking is optional)

A lot of news this week, but remember you control your home and classroom and what comes in it and out of it. Keep it simple, as my Army dad use to say. Play one round of Robot Says... Check It! If you're in Maryland, ask your district about its AI policy before the October deadline. And if there's an AI toy on a holiday wish list, have the "does it listen and remember what you say?" conversation now. Because toy companies know how to advertise and get your kid wanting the newest thing. Not to mention if there friend already has one.

This week we were all reminded to slow down, put people first, and teach kids to think and question AI before they trust it. That in my mind is AI literacy. And it can start at the kitchen table, not on a tablet. ;)

Thanks for being back here with me.

See you next Tuesday,

Amber Ivey (AI)