Weekly roundup

Here’s what you may have missed

Last week a stranger got a robot toy to say the Earth is flat. This week the kids pushed back. Nearly 100 students at a New York City arts high school walked out of class to demand a freeze on AI, and a big part of why was that adults keep handing student work to it. Meanwhile, an AI school that promises kids will learn twice as fast hit that mark for fewer than half of them. And kids are making fake explicit images of their classmates, here and overseas. That one is hard to read. But we need to really sit with that and read it anyway.

There's a new AI for Kids Podcast episode this week, and the timing is pretty good: Should Schools Ban AI? Vibe Coding, Cheating, and Keeping Your Brain in Charge. I talked with Dora Pálfi, co-founder and CEO of Imagi, about whether middle schoolers still need to learn to code and why you never take AI's first answer. Listen with your older kids, then ask them where AI should stay out of school.

This past weekend I flew out to the Bay Area for Khan Lab School's AI in Education Summit, where this year's theme was keeping learning human in an AI world. The event was great. While I was out there, we also set up a booth at New Kid Day in San Francisco, and introduced hundreds of kids and families to AI literacy with the ABCs of AI deck and sample pack. The feedback was great and it was so fulfilling to get this into the hands of families. Also so cool to see kids doing the activities IRL.

Got a few minutes? Here's the rest.

National Focused Article

An AI school promised kids would learn 2x faster. Fewer than half did.

The New York Times obtained internal records from Alpha School, the private school network where kids spend their mornings learning on AI-powered apps instead of with a teacher, and their afternoons in workshops. Only 42% of students grew twice as fast in both math and reading, the "2x" promise at the center of Alpha's marketing. Kids who started behind had a harder time. About a fifth of students scored at or below the 50th percentile in math last fall, and only 37% of that group hit 2x growth by winter. Parents and employees told the Times some students spend far longer than two hours a day on the software, and some families hire tutors. Tuition runs as high as $75,000 a year, and the network is opening campuses in Chicago, Atlanta, Charlotte, Raleigh and California. Many Alpha students still score well above national averages. Critics have also questioned how Alpha calculates "2x": if a student is expected to grow 7 points on a test and grows 14, Alpha counts that as 2x.

What parents and teachers need to know: "2x" is a marketing and this is what happens when somebody outside the company finally checks it. Some kids did great. But look at who got the least out of it: the kids who were already behind. Those are the kids who need a person the most. So when a school, a tutoring app, or a shiny new program says AI will help your kid learn faster, ask three things. Faster than what? Who measured it? And how did it work for the kids who were struggling?

Quote

“I really applaud and respect the students who say that, hey, I want to protect my own thinking, and so I'm not going to use this to fully offload.”

-Dora Pálfi, co-founder and CEO of Imagi, on this week's AI for Kids Podcast

Local Focused Article

NYC high schoolers walk out of class to protest AI

Nearly 100 students at LaGuardia High School walked out of class on October 20 and marched to Lincoln Center to demand a freeze on AI. LaGuardia is the Manhattan performing arts school that inspired "Fame." Students said AI had already shown up at their school: posters for school events, including a talent show, were made with Canva AI and ChatGPT instead of by student artists, and free materials from a national student-belonging campaign came with AI-generated music. Senior Mieke Nievergelt said the adults bringing AI into schools built their thinking skills before AI existed and don't understand what it does to young minds. The walkout was part of a National Walkout to Freeze AI organized by the group Irreplaceable, with rallies in about 50 cities and campuses. New York City paused student AI for pre-K through 8th grade this year, but high schools can run approved AI pilots and every high school student has to complete AI literacy training. Schools Chancellor Kamar Samuels has defended the different rules for high schoolers, saying AI is a growing part of the world they're about to enter.

What parents and teachers need to know: This was my favorite story this week. Grown-ups keep arguing about what AI will do to kids, and these kids answered for themselves. Look at what set them off. It wasn't a chatbot doing homework. It was adults using AI to make posters that student artists would have loved to make. So before you use AI for the class flyer or the fundraiser graphic, ask if there's a kid who wants that job. Then ask your own kids where they think AI belongs in school and where it doesn't. They have opinions and pulling them out helps them think critically.

Research Focused Article

Can an 8-year-old spot an AI video? Some can.

Brigham Young University researchers showed 8- and 9-year-olds a set of videos and asked which ones were made by AI, according to an October 22 write-up by BYU professor Sarah M. Coyne. Kids did better than guessing, but fewer than half got every video right. The 55% who said they knew what AI was did better than kids who didn't. Most kids judged by whether a video could really happen, and a video of babies riding animals in a fashion show was the easiest to catch. Some noticed strange lighting, skin or hair. Very few noticed the caption that said the video was AI-generated. The study is still under peer review.

What parents and teachers need to know: The kids who knew what AI was did better at spotting it. That's an argument for starting AI literacy early, right there. Start with the easiest check: could this really happen? Then look for weird hands, too-smooth skin, and voices that don't match lips. And point out the AI label when you see one. Most kids in this study skipped right past it, and a label only helps if somebody taught you to look for it.

AI Safety Article

Kids are making fake explicit images of classmates

The Denver Gazette reported that students in at least two Colorado schools made explicit AI-generated images of classmates this year and shared them, according to Safe2Tell, the state's anonymous tip line. In one case in August, Safe2Tell logged the report as blackmail. The tip line doesn't track AI misuse separately, so nobody knows the real number. Reports to the National Center for Missing & Exploited Children involving generative AI went from about 4,700 in 2023 to roughly 440,000 in the first half of 2025. A Common Sense Media survey found 44% of teens had seen AI-generated sexual material, and only 13% said they'd talk to a teacher or counselor about it. Colorado now makes creating and sharing these images a crime, with tougher penalties when they show minors. The same week, Malaysian police confirmed reports that AI nude images of more than 300 girls at one secondary school in Seremban were made from their faces, shared among students and allegedly sold on Telegram. A male student was suspended. Parents say the school kept them in the dark.

What parents and teachers need to know: I know this one is heavy. But these images take seconds to make with free apps, and most kids won't tell an adult when it happens. Supreet Mann at Common Sense suggests lots of short talks instead of one big one, and car rides are perfect because they have a built-in ending. Here's what I'd say: Making a fake picture of a real person is not a joke. It hurts people, and in a lot of places it's a crime. If you see one, don't share it, and tell a grown-up. Make sure your kids know they won't get in trouble for telling you. If it happens to your child, Take It Down can help get the images removed. Schools, make sure every student knows exactly who to go to.

Screen-Free Game: Prove the data!

This week's game comes from the Alpha School story: a big, catchy promise ("learn 2x faster!") that held up for fewer than half the kids once someone checked the numbers. Kids get to be the sellers and the testers.

What You Need: paper, markers, a timer, and a few everyday objects (a pencil, a sock, a spoon, a cereal box)

How to Play
  1. Make a product. Each player picks something from around the house, like a pencil, a sock or a spoon. Give it a big promise and write it on paper like an ad. Example: “This pencil makes you spell twice as fast!”

  2. Sell it. Take turns being the Seller. You get 30 seconds to tell everyone why your product is amazing.

  3. Ask questions. Everyone else is a Tester. Testers ask the Seller three questions:

    • “Twice as fast as what?”

    • “Who checked?”

    • “Did it work for everybody, or just some people?”

  4. Prove the data! If the Seller can’t answer, the Testers shout “Prove the data!” Now everyone plans a test together. Try it once with the product and once without, and keep everything else the same. Example: one person spells words for one minute with the “magic” pencil, then one minute with a regular pencil. Count the words both times.

  5. Run the test. Did the “magic” pencil really spell twice as many words? Try it with two or three more people. If it only worked for one person, the promise isn’t true for everybody.

  6. Score points. Sellers get 1 point for the funniest promise. Testers get 1 point for each question the Seller can’t answer. Everyone gets 1 point when the group plans a fair test.

  7. Closing question: “If an app said it would make you learn twice as fast, what would you want to know first?”

Why it works: In the real story, the promise sounded great and some kids did great. But when someone finally looked, it didn't hold up for most kids, and the kids who were behind got the least out of it. This game teaches kids that a number in an ad is not proof, that "faster than what?" is a real question, and that a fair test checks everyone, not just the kids who were already winning.

The voices on AI this week were teenagers.

I am so happy. We spend a lot of time talking about kids and AI. And talking to kids about AI. This week a group of teens reminded us to talk with them. So ask yours tonight: Should your school use AI? Where should it stay out? Their answers might change yours.

If this resonated, forward to a friend or colleague.

Until next week,

Amber Ivey (AI)

P.S. Has AI shown up at your kid's school in a way that surprised you? Hit reply and tell me.

P.P.S. Already making a holiday list? The ABCs of AI Activity Deck introduces kids to AI literacy with 130 cards and zero screens. No Wi-Fi, no microphone, no "2x" promises. ;)

Check out our recent show in San Francisco where we introduced hundreds of kids and families to AI literacy via our deck and sample pack.

AiDigiCards at Khan Academy School

Amber Ivey (AI)