We are back, and apparently September decided we needed a lot to talk about.

I took a longer summer break from the newsletter than I planned, but honestly, coming back now feels like good timing because something interesting is happening in the AI and kids conversation.

For a while, so much of the discussion was about getting AI into schools. Which tools should teachers use? How quickly should students learn them? How do we prepare kids for an AI-powered future?

This month, that changed.

Now some of the biggest school systems in the country are asking whether we moved too fast. Teachers are pushing technology companies for stronger protections. Child-safety organizations are testing the AI products kids already have access to. At the exact same time, hundreds of millions of dollars are being committed to putting even more AI into education.

AI is not disappearing from our kids' lives. But that does not mean every AI tool belongs in a classroom, every child needs a chatbot, or every use of AI counts as AI literacy.

Let’s get into what happened.

On September 2, New York City Public Schools announced a one-year moratorium on student-facing generative AI for children in 2-K through eighth grade. The policy affects nearly 600,000 students. High school students will still receive AI literacy instruction, and a limited number of high school classrooms will participate in supervised AI pilots. New York also prohibited AI companion chatbots across all grade levels.

That same day, Los Angeles Unified confirmed its own moratorium on students using generative AI on district-owned devices during the current school year. The district is not eliminating technology altogether. Students can still use tools such as speech recognition, text-to-speech, adaptive learning systems and translation technology. What the district is pausing is student access to generative AI systems that can produce answers, writing and other content for them.

These are the two largest public school districts in the United States.

What parents and teachers need to know: AI literacy and AI use are not the same thing. A five-year-old does not need a ChatGPT account to begin understanding that computers can recognize patterns. An eight-year-old can learn that AI makes mistakes without typing a prompt. A middle school student can learn how an algorithm influences what they see online before asking one to write their homework.

If your school says it is “teaching AI,” ask what that actually means. Are kids learning about AI? Are they learning with AI? Or is AI doing work for them?

Then the Gates Foundation announced that it plans to spend at least $1 billion over the next two years expanding access to AI and AI-enabled tools.

About 40% of that money is expected to support education, including AI tutoring for students and AI tools for teachers in the United States and around the world. That works out to roughly $400 million focused on education if the funding follows the announced proportions. That is a lot of money. And it is a reminder that even while some school districts are slowing student access, investment in AI for education is moving very quickly in the other direction.

I do not automatically think that is a bad thing. There are real possibilities here. An AI tutor might help a child practice a concept in another way. A teacher might use AI to adapt a lesson for different reading levels. A school system might use AI to make information easier for families to find.

What parents and teachers need to know: When someone introduces a new AI education tool, start with the learning problem, not the technology. What is this supposed to help a child do? What evidence shows that it works? Is it helping the child think, or simply making the task easier? And what role does the teacher still play?

Microsoft, the American Federation of Teachers and the United Federation of Teachers announced a new National AI Safety & Privacy Standard for schools. The agreement creates enforceable protections around Microsoft education products and gives school districts the ability to include those protections directly in their contracts.

Among the commitments, Microsoft says student and educator data will not be used to train AI models, sold or repurposed. Schools are supposed to maintain control over how their data is collected, retained and deleted. The agreement also requires human oversight for important decisions and calls for clearer explanations to families about how AI tools work and what information they collect.

What parents and teachers need to know: When your school introduces an AI tool, ask what happens to the information children put into it. “We protect student privacy” is not enough. Ask whether student conversations train the AI, how long the information is kept, whether a third party can access it and whether families can opt out.

Common Sense Media’s Youth AI Safety Institute released a new risk assessment of Perplexity, the AI-powered search tool. After testing its search and chat features, the Institute gave Perplexity its lowest rating for youth use: “Unacceptable Risk.” The assessment said researchers encountered problems involving age protections, potentially harmful content and inaccurate information. The organization tested about 1,200 searches before issuing its assessment.

This matters because AI search looks familiar. A child may not think, “I am talking to artificial intelligence.” They may simply think they are searching the internet. But traditional search and AI-generated answers are not the same thing. A search engine usually gives you places to look. Generative AI can take information from many places, combine it and hand back one confident-sounding response. That response can still be wrong. And kids are not always great at recognizing confidence without accuracy. To be fair, adults are not always great at that either.

What parents and teachers need to know: Start teaching kids to ask, “Where did that answer come from?” If an AI tool gives them a fact, especially about health, history, science or something important, show them how to open the sources and check.

With everything happening this month, this week’s activity is really simple.

Ask your child a question that does not have one obvious answer. You could ask, “How would you build a playground on the moon?” “How could we keep an ice cube from melting?” or “What would you invent to help people during a really hot summer?”

Before anyone searches online or asks an AI tool, give everyone two minutes to come up with an answer. There are no wrong ideas.

After everyone shares, talk about what you might ask AI to help with next. Maybe you want more ideas. Maybe you want to check whether something is scientifically possible. Maybe you want to learn about something you did not understand.

Focus on the order. Your brain first. AI second. Or brain before bot. (Check out our most recent podcast episode.)

One of the habits I want kids to develop is realizing that technology does not have to be the starting point for every question. Sometimes AI can expand your thinking. But first you need some thinking to expand.

Ask your school what its AI policy actually says. Ask what happens to student data when kids use AI tools. And the next time a child reaches for AI, Google or even you to answer something immediately, try asking one question first: “What do you think?”

They might be wrong. They might come up with something completely ridiculous. Good. That is part of learning too.

September’s biggest AI-for-kids stories are not really telling us to run toward AI or run away from it. They are reminding us that the adults responsible for kids need to be much more intentional about where AI belongs. Because AI literacy is bigger than knowing how to prompt a chatbot. Kids need to understand what these systems are doing, when they can help, when they can get things wrong and when their own brain needs to do the work first. And yes, I still believe that can start without a screen.

Thanks for being here and taking this seriously.

See you next Tuesday,

Amber Ivey (AI)