Schools Are Sitting Ducks
Gaggle, the AI surveillance company that schools use to monitor student writing and behavior, exposed 3,500 sensitive student records in Vancouver Public Schools last year. Poems. Essays. AI chat transcripts. All accessible through unprotected links. No password. No encryption. No access controls.
That’s not a worst-case scenario. That’s what happened when a company with access to thousands of schools decided that basic security wasn’t worth the engineering effort.
Six months before that, a college kid in Massachusetts exploited basic security gaps in PowerSchool’s customer support portal and got access to 62 million student records — the largest data breach of children’s data in U.S. history. Not through sophisticated hacking. Through a gap that shouldn’t exist.
Schools aren’t targets because they’re fortress-like. They’re targets because they’re soft. K-12 schools face 4,388 cyberattacks per week. Eighty-two percent of K-12 schools experienced a cybersecurity incident between July 2023 and December 2024. That’s not outliers. That’s normal.
And now schools are adding AI tools on top of infrastructure that was already struggling.
Why Schools Are Unprepared
Most school districts don’t have real IT departments. They have one person. Maybe two. That person is managing 1,500 devices, handling password resets, patching software, and now they’re supposed to audit the security of the AI tool the superintendent just approved because it’ll save 10 hours a week on grading.
It’s not that schools are careless. It’s that they’re understaffed and under-resourced. A district in rural Montana doesn’t have a cybersecurity officer. They have a tech guy who’s also running the WiFi and fixing the printers.
The other problem is simpler: nobody’s forcing AI companies to secure their products before they sell them to schools. Gaggle didn’t secure its database because there was no reason to. No lawsuit waiting. No regulation. No consequence. So why spend the money?
The AI Company Problem
Here’s what should happen: before an AI tool gets access to student data, it should meet a security standard. It shouldn’t. It doesn’t.
Anthropic, OpenAI, Google — they all publish responsible AI principles. They all talk about safety. But the moment those tools hit the market and schools buy them on a three-year contract, the company’s incentive is to ship fast and patch later. The cost of a breach is a PR problem and maybe a settlement. The cost of delaying launch is lost revenue.
And when you stack AI agents and “swarms” on top of that — tools that are supposed to automate decisions and access student data at scale — you’re multiplying the exposure.
What Schools Can Actually Do
You can’t wait for AI companies to police themselves. You can’t wait for Congress. Monday morning, do this.
First: any tool that touches student data needs a security audit before it gets access to your network. This means your IT department sits down with the vendor and asks: where’s your data stored? Who can access it? How is it encrypted? What’s your incident response plan? If they can’t answer, the tool doesn’t get installed.
Second: basic controls. Password-protect your data. Encrypt it. Control who can access it. This sounds obvious because it is. Gaggle had none of it. PowerSchool’s support portal had minimal controls. Your school should have more.
Third: assume something will break. Not if. When. You need a plan for when a vendor gets breached. Where does your data get restored from? How do you notify families? What’s your communication strategy? Most schools don’t have one.
Fourth: your IT person is drowning. You know this. Districts need to fund information security the way they fund building maintenance. It’s not a luxury. It’s infrastructure.
The Bottom Line
Schools aren’t secure because they’re not built to be, and AI companies don’t have incentive to make them secure. That’s a gap that only closes when schools demand security before they buy, and when someone else is hurt badly enough that regulation finally kicks in.
Until then, yes — schools are sitting ducks. The Gaggle breach and PowerSchool breach weren’t anomalies. They were previews.
If your district is evaluating a new AI tool, that’s your moment. Don’t just ask if it’ll save time. Ask if it’s locked down. If the vendor can’t promise basic security, the answer is no — until they can.
If you’re thinking through how to protect student data while building a school culture that uses AI thoughtfully, my book goes deeper into the real questions. The AI Doesn’t Know Your Students is available on Amazon and at shouldiuse.ai/book.
