After our fall break, we're back, and we're writing this one from NY EdTech Week. If you're here too, come say hi. We'd love to hear what you're seeing in your district this fall.
Okay, so let's get into it. Frederick County Public Schools in Maryland followed the guidance documents. It evaluated AI tools before putting any in front of students, and its curriculum director said Google's Gemini was the only one that met the state's safety and privacy standards. This fall it let middle and high schoolers use Gemini with teacher permission, while blocking other AI platforms on school devices.
Fast-forward a few weeks, and the board voted 6-1 to ban student AI use from pre-K through 12th grade. Board members said Gemini was turned on without warning parents, and educators raised concerns that teachers couldn't reliably turn it off on student devices. Teachers and the student board member pushed the ban past the younger grades the board first considered. The board allows exceptions for AI literacy lessons and disability accommodations, and the president described the ban as temporary, "not forever," while the district builds a pre-K through 12 curriculum on proper use.
Frederick isn't alone. Los Angeles blocked generative AI on every student device in September, a move that surprised some of its own board members, and (as we wrote about earlier) New York City paused student AI through eighth grade the same week.
Also, on September 30, the Wall Street Journal published an investigation, based on more than 60 interviews with teachers, officials, parents, students, and clinicians, into students becoming academically dependent on Google's classroom AI. It's definitely worth a read.
Reading the coverage side by side, two problems keep showing up. Districts couldn't control the AI once it arrived, and parents couldn't see it until something went wrong. Both are solvable problems.
Some folks have reached out to us asking how to subscribe themselves or their colleagues to this newsletter. Click here to do so. We send a newsletter every two weeks about the latest happenings with K-12 and AI.
IN THIS ISSUE:
Why this fall's bans are what happens without a layer between the district and the AI companies
Why parents need to see AI in their kids' schools before it becomes a board meeting
6-1
The Frederick County (MD) board vote to ban student AI from pre-K through 12th grade, weeks after piloting Gemini. Source
378,000
Los Angeles students now blocked from generative AI on district-issued devices. Source
20 pts
How much lower daily AI users who relied on it to draft writing scored in science on the 2025 PISA exam, compared to students who never used AI. That's roughly a year of schooling. Source
1+year
Math progress students gained in eight weeks using Gemini's Guided Learning mode, which coaches instead of answering, in a Google-funded trial. Same company, different design. Source
60+
Teachers, officials, parents, students, and clinicians interviewed for the Journal's investigation. Source
Boy, oh boy, is Gemini in the news. In Frederick County, the district turned on Gemini for students before parents knew much about it, and concerns focused on what teachers couldn't control once it was in use. In Los Angeles, administrators blocked AI (including Gemini in Google Classroom) on student devices before the board's new AI committee had met, according to LAist. And in Boston, a former chief information officer summed up the problem to the Wall Street Journal: "You cannot escape Gemini in Google Classroom."
When AI comes straight from the company that built the model, inside tools a district already depends on, the district's options shrink to two: leave it on for everyone or turn it off for everyone. That's a particularly tough spot to govern from. Data in that same Journal piece shows why an on/off switch is the wrong tool. On the 2025 PISA exam, daily AI users who relied on it to draft writing scored about 20 points lower in science versus students who never used AI, while daily users who used it to help them learn and had AI literacy training scored slightly higher. Conversely, students using Gemini's Guided Learning mode (according to a study commissioned by Google itself, so decide how much you trust that) gained more than a year of math progress in eight weeks, with the AI coaching instead of answering.
Bottom line: Student outcomes depended on how the AI was set up and used.
OUR TWO CENTS
I don't read Frederick County's vote as a verdict on AI, per se. I read it as what a board does when its only options are to “leave it on” or “turn it off”. If you can't allow a tool for sophomores while blocking it for second graders, set it to coach instead of answer, or control when new features show up, then every decision is all or nothing. All or nothing usually ends at “nothing”, and honestly, I'd probably vote the same way if I were them. All or nothing isn't great.
Last month, we wrote about why districts need a layer between them and big AI models, one that answers to the district, not the AI company. The news over the last few weeks made the case even better than we did. A governance layer is where the dials and switches live: which grades get which tools, whether a tool helps outline a paper instead of writing it wholesale, if and when new features reach students, and the ability to swap models if one starts misbehaving. You should have dials and controls, not just 'take it' or 'leave it'. If the answer is "it's on, or it's off," Frederick County's Board has already shown how the next AI vote will go.
List every place students can reach AI on district devices, including features built into tools you already license, not just apps you approved.
For each one, find out what you can control: by grade, by feature, by classroom, and whether it tutors or answers. Every gap is a vendor conversation.
Ask vendors how much notice you'll get before new AI features reach students, and get it in writing.
If your district pauses AI, write down at the same meeting what would have to be true to turn it back on.
Notice how often parents show up in this fall's stories as the last to know. Frederick's board members said Gemini reached students without much warning to families. In Los Angeles, some board members learned at a committee meeting that AI had been blocked. In Brooklyn, the Journal reports, parents pressed officials at an August education council meeting after learning consumer chatbots had been fully unblocked on their children's school devices the previous fall. Meanwhile, families are compelled to ask about AI: The Washington Post ran a back-to-school guide in September telling parents which questions to bring to their child's school about AI tools.
Parents who can't see what's happening fill the gap with the worst version they've read, and this fall there's been plenty to read. When parents aren't consulted and can't see how AI is used, a perfectly reasonable pilot can become a complete ban in just a few weeks. Visibility means families know which AI tools their child can use, what those tools are set up to do, what happens to their child's data, and how to raise a concern before it turns into public comment.
OUR TWO CENTS
Like everything PR-related in K-12, getting ahead of this is 90% of the work. Publish answers before the questions arrive. Set up a page on the district website, updated whenever something changes: the AI tools students can use, by grade; what each one is set to do (tutor, help with writing, off); what student data goes to it and whether it's used to train anything; who to contact with a concern; and how to opt your child out. Then give parents notice before a new AI tool or feature reaches their kids, the same way you'd notify them about a field trip or a new curriculum. Put it in your school's newsletter. For younger students, consider going further and giving families a simple summary of how their child is using AI. None of this requires state guidance, some new law, or a budget revision. It requires deciding that parents hear about AI from the district first. If the layer Russ described is where the switches live, this page is where the district shows families how they're set.
Publish a one-page AI page for families: tools by grade, what each is set to do, what data goes where, and who to contact.
Commit to notifying parents before any new AI tool or feature reaches students.
Offer an opt-out with an alternative assignment, so choosing it doesn't cost a child a grade.
For younger grades, explore giving parents a simple view of how their child uses AI at school.
With Parents Divided on Artificial Intelligence, Here's How Schools Can Build Trust, EdTech Magazine
This piece starts from a statewide EdTrust poll of more than 1,300 Massachusetts parents, who split roughly into thirds on AI in the classroom: positive, negative, and unsure. From there, the experts it interviews explain what moves that unsure third, including clear policies, plain communication about privacy, and showing families that the district is actively defining responsible AI use. It's a quick read and a good one to share with whoever handles family communications in your district.
Chat with ClassCloud
We’re listening. Let’s Talk!
This newsletter works best when it’s a conversation, not a broadcast. If you want to talk through how any of this applies to your district specifically—or if you have feedback on what would make this more helpful—just hit reply. We read and respond to everything.
Schedule a Virtual Meeting
Thanks for reading,
Russ Davis, Founder & CEO, ClassCloud ([email protected])
Tracy Walters, Director of Customer Growth ([email protected])
ClassCloud is an AI company, so naturally, we use AI to polish our content.
