AI and kids is all over the place. Congress is talking about whether schools can use existing federal dollars for AI literacy. Europe is reminding schools that teachers still need to be at the center. OpenAI is making safety updates to ChatGPT. And new research says a lot of teens are already having weird, uncomfortable, or harmful experiences with chatbots. We knew this, but the kids are saying it too.

So yes, the AI conversation feels like it is getting louder and more concerning. But for parents and teachers, your questions can stay at level one: what are kids using, what is it doing, and are we actually talking to them about it?

If you missed last week's podcast, catch up here.

Here is what is in front of us this week.

Rep. Randy Fine (R-Fla.) introduced the K–12 AI Literacy and Readiness Act of 2026, which would amend the Elementary and Secondary Education Act of 1965 to allow states and school districts to use existing federal education funds for AI instruction and teacher training. The bill does not create new spending; it clarifies that current allowable uses include AI literacy at both the state and local level. Fine framed the proposal as a response to schools that, in his words, lack the authority or resources to keep up with the pace of the technology. The bill has been endorsed by the Computer and Communications Industry Association, which described AI literacy as a core workforce skill. It joins a wave of state-level activity: more than 50 AI-in-education bills are now moving through 25 state legislatures, with provisions ranging from parental opt-in requirements to bans on AI making high-stakes decisions about students.

What parents and teachers need to know: If you are a parent: this is a moment to ask your district what their AI plan actually is. Not the policy memo, the actual plan. Who is teaching it, what is being taught, and how is it assessed? If you are an educator: the three-question framework still works as a starting point even for you all as AI comes to your school. Who made this tool? Why did they make it? Why is this platform showing it to me?

The Council of the European Union approved conclusions calling for an ethical, safe, and human-centered approach to AI in education, with teachers held firmly at the center of the learning process. Dr. Athena Michaelidou, the Cyprus minister for education, described teachers as “guides, mentors and critical thinkers” rather than mere users of AI tools. The conclusions emphasize digital skills, AI literacy, inclusion, fairness, and the wellbeing of both teachers and learners. The Council acknowledged AI’s potential to improve accessibility, individualize instruction, and reduce administrative burden, while flagging concerns about reduced student autonomy, over-reliance on technology, bias, misinformation, and data protection. The conclusions specifically warn that AI could deepen existing inequalities and digital divides if deployed without care.

What parents and teachers need to know: “Human-centered” is a phrase that means different things to different people. When you hear it from your school, ask: human-centered for whom? The teacher? The student? The administrator running the dashboard? Teachers being at the center of learning is essential. Kids learn through relationships, and a relationship with a chatbot is not the same kind of relationship. The EU framework is worth knowing about even if you live elsewhere. European rules tend to set the floor that global tech companies eventually build to.

OpenAI announced updates to ChatGPT designed to recognize when risk is emerging across the course of a conversation. not just within a single message. The mechanism, called “safety summaries,” produces short factual notes about earlier safety-relevant context and feeds them back into the model when a current request raises concern. In long single-conversation tests, OpenAI reported safe-response rates improving by 50 percent for suicide and self-harm scenarios and 16 percent for harm-to-others cases. On GPT-5.5 Instant, the company’s current default model, the same updates improved safe-response performance by 39 percent and 52 percent respectively. The update was developed with input from psychiatrists and psychologists in OpenAI’s Global Physicians Network.

What parents and teachers need to know: Improvements in safety measured against the company’s own benchmarks are not the same as independent verification. However a 50 percent improvement is significant. If a teen in your life uses ChatGPT, this update applies to them automatically, there is nothing they need to turn on. But also nothing they can turn off if they prefer a less monitored conversation. A good place to have a conversation with the kids in your life. Keep in mind, these changes were announced the same week new lawsuits were filed.

A new study from the Cyberbullying Research Center looked at 3,466 U.S. teens ages 13 to 17 and found that many teens are having troubling experiences with AI chatbots. Among teens who use conversational AI chatbots, almost half said they had at least one harmful experience. About one-third said a chatbot asked for personal information in a way that made them uncomfortable. Others said a chatbot pressured or manipulated them, encouraged them to do something wrong or illegal, or even encouraged self-harm. The findings were especially concerning for 13-year-olds, the youngest teens in the study. They were more likely than older teens to report being encouraged toward illegal behavior, self-harm, or suicidal thoughts. Male teens also reported these harms at about twice the rate of female teens. The researchers, Sameer Hinduja and Justin Patchin, said this study is important because it gives real data on teen chatbot use, instead of relying mostly on stories and individual examples.

What parents and teachers need to know: Researchers are clear that when kids think telling the truth will get a tool taken away, they stop talking. Thirteen is still very young, and kids that age are still building impulse control, judgment, and resistance to pressure. Instead of asking, “Is anything bad happening?” try, “Tell me about the weirdest or most uncomfortable thing a chatbot has ever said to you.” Boys reporting harm at twice the rate of girls is also a signal that we may need to talk to boys about AI in a more direct way.

This week’s game introduces the idea of chatbot memory, what it stores, what it forgets, and what it might share.

What You Need: a stack of index cards or scrap paper, one marker per player, and a small box, hat, or bowl labeled “Bot Memory”

How to Play

Why it works: Real chatbots keep information too. This game is about consent and memory, two things AI companies talk about a lot but kids rarely get to experience hands-on. Let the game get a little chaotic and surface the learning lessons as you play.

Adults cannot be completely hands-off with AI. This study is really concerning, and it reminds all of us how important it is to stay close to this topic. That does not mean we panic. It does not mean we ban everything. And it definitely does not mean we pretend kids are not already curious or using these tools. I think it does mean we ask better questions and we stay close enough to notice what is changing.

Also, please play the game this week. “What Does the Bot Remember?” looks simple, but it opens up a conversation many families and classrooms have not had yet. What are kids telling chatbots? What do they wish they had not shared? What do they think the bot does with all of it?

Please email me and let me know how it goes. I’d also love to hear what else you’re learning from the kids in your life. That helps me make this better.

See you next Tuesday.

Amber Ivey (AI)