Google Turned On AI for Every Elementary Student

This is an update to the elementary AI tools roundup published on this site earlier this year. Two things have changed since then that are significant enough to need their own coverage rather than a quiet edit, and both correct an assumption that most teachers, reasonably, still hold.

The assumption is that AI access for young students requires someone to actively turn it on. That assumption is no longer accurate for at least one major platform, and a real safety incident from last winter is worth understanding clearly before you assume every AI tool your school uses has been vetted the way you would want.

Google Switched Gemini On by Default for Every Age

On August 10, 2026, Google changed Gemini in Google Classroom from an opt in feature to a default on feature, meaning students of every age, including elementary students, now have Gemini enabled in their school Google account unless a school or district has specifically disabled it.

This matters because the previous assumption, reasonable at the time the earlier version of this piece was written, was that a school or teacher had to make a deliberate decision to introduce AI chat access to young students. That is no longer how it works. If your school uses Google Workspace for Education and has not specifically reviewed and adjusted this setting since August, your elementary students may currently have access to a general AI chat assistant by default, without anyone at your school having made an active decision that this was appropriate for their age group.

Check this directly rather than assuming either way. Ask your school's IT administrator whether Gemini access has been reviewed or restricted for elementary accounts since the August default change. This is a five minute conversation that resolves a real uncertainty, and it is worth having regardless of your personal view on whether young students should have this access, because the point right now is simply knowing what is actually turned on.

What Happened in Los Angeles and Why California Changed Its Guidance

Last December, a Los Angeles elementary school assignment asking fourth graders to create a book report cover led to a specific problem. A student used a school provided Adobe AI image tool to generate cover art for a book report on Pippi Longstocking, and the tool produced inappropriate imagery. Parents who heard about the incident tested the same tool themselves and were able to replicate the problem, which led to public concern and, within weeks, updated AI guidance from the state of California.

This incident is worth understanding specifically, not as a reason to avoid AI tools broadly, but as a concrete illustration of a real risk with a real cause. Image generation tools, even ones marketed and provided through a school licensed platform, do not universally have the same content safeguards for young user contexts as tools built specifically for education from the ground up. A general purpose creative AI tool licensed to a school is not automatically the same thing as an education specific AI tool with age appropriate content filtering built in as a first principle rather than an add on.

The practical lesson for elementary teachers: before using any AI image generation tool with students, including ones your school already licenses for general use, verify specifically whether it has documented, education specific content safeguards, rather than assuming a school license implies that verification has already happened. This is a genuinely different question from whether a tool is generally reputable, and the Los Angeles incident shows the gap between those two things clearly.

The Access Landscape Is More Fragmented Than It Looked in May

Independent tracking of AI tools used with elementary aged students has found that out of ninety tools reviewed in one directory, only thirty state a minimum user age below thirteen or no minimum age at all, and only fifteen of those thirty explicitly declare COPPA compliance. Just three tools in that same review, Knowt, Kami, and EnlightenAI, hold an independent third party security assessment rather than relying solely on the vendor's own claims.

This is a meaningfully more fragmented picture than a simple list of recommended tools suggests. The deciding question worth asking about any tool before using it with elementary students, more useful than a general reputation check, is whether the teacher can see every message or interaction the child has with the tool in real time. Tools built around full teacher visibility, rather than private student to AI conversation, address the core safety concern directly regardless of the specific platform.

SchoolAI, covered previously on this site as a strong option specifically because of its teacher visibility model, is worth a correction here. Current reporting places its free plan at a cap of five Space sessions per year, which is considerably more restrictive than casual classroom use would suggest and worth knowing clearly before planning a workflow around free tier access.

What Genuinely Improved

Not everything in this update is a caution. A few tools have matured in ways worth knowing about specifically for elementary use.

Snorkl, designed for grades one through five, lets students explain their math thinking by voice and drawing rather than typing, which fits how young students actually think through a problem far better than a text based interface does. Kami's read aloud and annotation tools are specifically noted as not being open ended AI, meaning there is no chat interface for a student to wander into unexpected territory with, and the tool is reported to collect no student data, which is a meaningfully cleaner privacy position than most AI adjacent tools in this space.

What This Means for Your Classroom This Term

Check your school's current Gemini in Classroom setting for elementary accounts specifically, given the August default change. Do not assume a school licensed creative or image generation tool has education specific content safeguards without verifying that directly, particularly in light of the Los Angeles incident. Prioritize tools with full teacher visibility into student interactions over tools promising a private, unsupervised AI conversation, regardless of how that tool is marketed. And verify current free tier limits directly before building a workflow around any specific platform, since the actual limits, as with SchoolAI's five session cap, are sometimes considerably more restrictive than general coverage suggests.

None of this means avoiding AI tools in elementary classrooms. It means treating the specific safety and access questions with the same seriousness this site has applied to special education documentation and behavior tracking, because the population involved here, children under thirteen, carries the same weight of care.