Last week felt like we were drinking from a fire hydrant. Thankfully, this week slowed down a bit. More stream than storm.

That is where we are starting this week.

A computer helped doctors solve medical mysteries, but doctors still had to make the call. Senators talked about AI in classrooms, but the conversation kept coming back to student safety, teacher judgment, and what kids actually need. And researchers reminded us that some of the “smart” toys marketed to families are doing more than playing along. They may also be collecting information.

But before we jump in, the Season 3 finale of the AI for Kids podcast is out this week. We are wrapping with the lessons I want kids to take into summer: question what you see, slow down when something feels too easy, and remember that AI is just a tool. Your brain still matters.

Listen here: Your Brain Still Matters + Season 3 Recap and Finale (Kids, parents, & teachers)

Okay, let’s get into what happened this week.

Researchers from Boston Children's Hospital and OpenAI announced that off-the-shelf AI tools helped identify the genetic errors behind a set of hard-to-solve childhood illnesses. Published in the New England Journal of Medicine's AI publication, the findings show that OpenAI's o3 Deep Research model helped clarify 18 diagnoses for children who had struggled for years to find a cause for their symptoms. One researcher said she was shocked a commercial system could spot new diagnoses in genomes that had already been analyzed many times, work that can take human analysts days per case.

What parents and teachers need to know: This is the kind of story worth telling your kids, because the headlines they usually hear about AI are scary ones. Here's the balance to pass along: the AI was fast at finding patterns in a giant pile of information, but human doctors still had to confirm what was real and decide what to do.

A U.S. Senate subcommittee held a hearing on the role of AI in K-12 schools, where Delaware's education secretary and other specialists testified. Education specialists urged lawmakers to ease the federal regulatory burden on AI in schools, arguing it could give teachers a powerful aid and help students learn the technology shaping their future. At the same time, the hearing featured a call for AI that protects "human judgment" in classrooms, a reminder that "use it" and "use it carefully" were both on the table.

What parents and teachers need to know: When the people writing the rules are still debating the rules, that tells us that NOBODY has the final answer yet. Not Washington. Not the experts. Not the companies building the tools. So don’t wait. Ask your school district now: When AI shows up in my kid’s classroom this fall, what is the plan for keeping a teacher’s judgment at the center? You are allowed to ask before the tools arrive. Not after something goes wrong.

A June 17 analysis looked at how AI is moving into schools around the world, and how much the results depend on the design around the tool, not just the tool itself. It highlighted Germany, where the state of Bremen became the first to roll out a specially built classroom AI, called "Telli," across all its public schools for lesson prep and in-class use. The piece's core argument: AI can personalize learning and lighten a teacher's load, but it can also weaken core skills if it's dropped in without a plan.

What parents and teachers need to know: It helps to know we're not the only country figuring this out in real time. The lesson from the places moving fastest is the tool matters less than the thinking around it. A purpose-built school AI with trained teachers is a very different thing from a free chatbot a kid opens alone at midnight. When you hear "our school is using AI," a useful follow-up goes beyond "which one?" to it's "built for what, and with what training?"

Security researchers at Cybernews reported on the data habits behind popular AI and robotic toys for kids. They analyzed 10 Android companion apps tied to these toys and found that all of them asked for permissions Android classifies as "dangerous," and that 7 of the 10 contained third-party trackers. The researchers also flagged advertising, profiling, and location trackers, warning that these always-listening toys can harvest sensitive data with ease.

What parents and teachers need to know: A toy that talks back is also a toy that listens, and sometimes a toy that remembers and shares. Before one of these comes into your home, treat it like any other device, not like a stuffed animal: check what the app asks permission for, keep play in shared spaces rather than the bedroom, and turn it off when playtime's over. A teddy bear shouldn't need your location. If it's asking, that's your sign to ask why.

This week's game comes straight from the rare-disease story: AI is great at spotting patterns in a big pile of clues, but a human has to check the work before anyone trusts it.

What You Need: paper and a marker (optional)

Speed and correctness are not the same. The Pattern Machine sounds confident even when it's wrong, exactly like real AI. Once a kid has been the Checker a few times, "let me double-check that" starts being a habit.

Play one round of Pattern Detective. Ask your school or district what the AI plan is for the fall. Or just use one of these stories to start a real conversation with a kid in you life. They are already seeing AI in school, online, in games, and even in toys.

That is why I keep saying AI literacy has to start earlier than we think, and it does not have to start with a screen.

Thanks for being here and taking this seriously.

See you next Tuesday,

Amber Ivey (AI)