By Emily Cherkin, M.Ed., and Dina Alexander, M.S.
This spring, the Utah State Board of Education signed a data privacy agreement with Google, opening the door for Gemini, Google’s AI chatbot, to be used for instruction in K–12 classrooms across the state. Individual school districts decide whether to adopt it. Parental consent is not required. Students are enrolled by default.
No public vote was taken. No parent input was sought. And the most pressing question has not been answered: is this product actually safe for our children?
Typically, a data privacy agreement settles a compliance question. In this case, the DPA tells us Google has met Utah’s legal requirements for handling student data.
But it does not tell us whether Gemini is developmentally appropriate, educationally effective, or free from harm for children.
These are entirely different questions, and they remain unanswered. We demand answers.
What Google’s own research won’t tell you
We have spent years watching EdTech companies package and market their products for schools and policymakers in the name of “efficiency” and “alignment” and “equity” and then present them to leaders as one-size-fits-all solutions in spite of the lack of actual evidence. We can recognize this kind of claim the moment I see it.
Google recently published a study claiming its “Guided Learning” feature helps students think more deeply by asking probing questions rather than handing over direct answers. Perhaps it sounds promising on the surface. But the study was conducted on students in Sierra Leone, co-written with a Google partner company, and published on Google’s own website. This is not independent research. This is marketing with footnotes.
The effect size Google claims as “significant” falls well below the threshold researchers identify as meaningful in education (in education research, “meaningful impact” actually begins at standard deviations of approximately 0.40 – 0.50).
But the real problem goes deeper than statistics.
Educator Joe Clement tested the tool firsthand and found that Gemini abandoned its Socratic approach the moment he rephrased a question. Instead of engaging the student in thought-provoking questions first, it provided the answer. Immediately. As Joe put it: “Almost any kid is going to stop reading after being given the answer.” Anyone who thinks children are not savvy enough to figure this out is utterly unfamiliar with kids and willfully ignorant about how they actually use technology,” he observed.
Children already know how to use what Joe refers to as “reverse sycophancy.” Act interested in what the AI is selling, and it will do whatever you want. Students quickly figure this out on their own. The idea that a tool this easy to game will build critical thinking, grit, or deep understanding in a developing mind is not just optimistic. It is wishful thinking dressed up as innovation.
This matters because when children use a tool to complete a task before they have built the skills to do it themselves, they are not accelerating their learning. They are bypassing it entirely. The friction is not the obstacle. The friction is the point. We cannot hand children devices on which to shortcut the very experiences that build capable, independent minds and call it progress.
What Google’s own documents reveal
Internal Google records, as reported by NBC News, describe how Google sees schools as a “pipeline of future users” and discuss the business value of building brand loyalty through early exposure to Google products. Because to Google, EdTech products are not actually tools; they are sources of potential future revenue. Thus parents deserve to know that when a school adopts Gemini, they are not choosing an educational tool– they are enrolling their child in a customer development program.
Two active lawsuits against Google deepen our concerns. One is a wrongful death case, alleging Gemini’s chatbot contributed to a user’s suicide. Another is a proposed class action alleging Google activated AI features across its platforms without adequate user consent. Both cases are unresolved, but they are a clear warning that these products are unvetted, untested, and unsafe, and they are being deployed in classrooms by default.
Three things we are asking for.
The USBE has framed this rollout as though parent opt-out is sufficient protection.
It is not.
An opt-out framework places the burden on parents, most of whom have no idea what Gemini is, what data it collects, or what neurological trade-offs accompany its use. Opting out requires parents to navigate a bureaucratic process to protect their children from a default their school board never should have implemented. It is a consent framework designed to minimize resistance, not maximize protection. Opting out of EdTech is something Emily advocates for as a means to an end– not the end result itself– because it engages parents in the process of resistance and autonomy. But even providing an opt out pathway is not enough.
We do not apply opt-out logic to other risky educational decisions. We do not say, “We’ve decided your child will be in an unventilated classroom, but you can opt out.” We set standards that protect all children first. The same standard belongs here.
We are asking Utah to press pause on deploying GenAI products in schools.. Specially, we are asking for three things:
Informed parental consent must be obtained before children useGenAI products for instruction. Opt-in, not opt-out, should be the standard, and parents should never be forced to choose between protecting their child from a technology they have concerns about and ensuring their child still receives a meaningful education.
Before any GenAI product is approved for statewide use, there should be public notice, a real chance for parents and concerned voters to weigh in, and a vote by the elected members of the Utah State Board of Education. Decisions affecting hundreds of thousands of children should not be made quietly.
Utah families deserve a public registry or publicly posted list on school district websites documenting which EdTech and GenAI products have been approved for school use, who makes them, what they do, and what protections are in place (example). Transparency is not a burden. It is a baseline.
Utah has long recognized that parents are the primary decision-makers in their children’s education. That principle should not disappear the moment a technology company arrives with a product and a privacy agreement.
If GenAI products are optional for school districts, they should be optional for students too.
Add your name to the Child First Policy Center’s open letter to the Utah State Board of Education. Share this with your school principal, your PTA, and your state board representative. GenAI products in education is not a foregone conclusion. The window to shape this is still open.
Emily Cherkin is an educator, author, and nationally recognized expert on children and screentime. She writes at First Fish Chronicles on Substack.
Child First Policy Center is a nonpartisan, Utah-based child safety policy and advocacy organization.
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.