Magic Fountain in your district

What this is

A teacher-facing authoring tool for classroom instruction. A teacher enters the lesson topic they are teaching and the tool produces a review activity aligned to it — a Jeopardy-style board, a trivia round, an escape-room style problem set, or a scenario-based exercise. The teacher then runs that activity with their class, typically projected on a classroom display.

It is an instructional product used during lessons, not an entertainment gaming site. There is no advertising, no messaging or chat between users, and nothing to purchase inside an activity.

Shared activity library. Teachers may optionally publish an activity they created so other teachers can find and reuse it. This is opt-in — activities are private when saved, and publishing takes a deliberate action. Only signed-in teacher accounts can publish; students cannot post anything. Shared material is limited to structured lesson content — questions, answers, and activity prompts — generated under the enforced K-12 content policy described below. It is not free-form posting, and there is no comment or discussion feature anywhere in the product.

How AI is used

We want to be precise about this, because “AI” covers a wide range of tools and many districts have policies that distinguish between them. Magic Fountain has two kinds of activity and they fall on different sides of that line. Districts evaluating us should assess them separately.

Pre-generated activities — Trivia, Jeopardy, and Branching Story

AI runs once, before the lesson, when a signed-in teacher generates the activity. What it produces is fixed content the teacher reviews: a set of questions and answers, or in the case of a Branching Story, every scene, every choice, and every ending written out in full before anyone plays it. During play, students choose from options that already exist. Quiz scoring is a direct comparison against the stored answer key. No AI model is called at any point while students are working, and no student input is ever sent to a model. These are teacher-authored materials that happen to have been drafted with AI assistance.

Narrative activities — AI Escape Room and RPG

These are different, and we would rather be direct about it than let you discover it later. AI runs during play, in response to student actions. A student picks a suggested action or types one in their own words, and the system generates the next scene as text with an accompanying illustration. If your district restricts student-facing generative AI, these two modes are within the scope of that restriction and the quiz activities above are not.

Where they are permitted, they are bounded rather than open-ended. A student is never talking to a general-purpose chatbot. Every play session carries a signed token pinning the topic, activity type, and reading level the teacher selected, so an activity cannot be steered away from the lesson. Sessions are capped at 25 moves, each shared activity has a daily ceiling, students create no account, and no personal information is attached to a session.

Both kinds of generation are constrained by an enforced content policy written for K-12 classrooms, prohibiting sexual content, graphic violence, substance references, profanity and slurs, self-harm content, and material intended to frighten children. Topic requests are screened before generation, and a teacher always reviews an activity before using it with a class.

What we never do. AI is not used to grade student work, monitor student behavior, inform placement, promotion, or graduation decisions, develop special education plans, or respond to student mental health. Quiz scoring is a deterministic comparison against the teacher-reviewed answer key, never a model judgment. We do not train models on student data, and we do not sell or share it.

New York City Public Schools — 2026–27 AI and screen time policy

NYCPS has placed a moratorium on student-facing generative AI in grades 2K–8 for the 2026–27 school year, limited grades 9–12 to approved vetted programs, and set screen time guidance by grade band. The policy is specific about what it covers: “Software that uses student-facing generative AI is not permitted in grades 2K–8.”

Our activities are not all the same in this respect, so here is each one against that test.

Why the first four fall outside it

In each of them the AI ran once, before the lesson, when the teacher built the activity. What the student opens is fixed content the teacher has already read. No model is called while a student is working and no student input is sent to one, so there is no student-facing generative AI to restrict. They are teacher-authored materials that were drafted with AI assistance, in the same sense as a worksheet a teacher wrote with help from a tool.

Escape Room and Role-Playing Adventure are turned off for NYCPS

These two generate each scene during play, in response to what the student does. They fall under the 2K–8 moratorium, and in grades 9–12 they would require the approved-pilot route, which we are not part of. There is therefore no NYCPS grade at which they are permitted — so they are disabled outright for NYCPS accounts, at every grade level.

This is enforced in the product, not left to the teacher. Accounts identified as NYCPS do not see these activity types in the menu, and the server refuses to generate or launch them regardless of what a client asks for. The restriction applies to activities created before the policy as well as new ones.

Teacher-led whole-class use

The policy states that “teacher-led use of tools—such as computers, tablets, and interactive smart boards—is allowable for whole or small group instruction.” The Jeopardy board and the printable review sheet are built for exactly that: the teacher runs them from the front of the room and students need no device at all, so the 1:1 screen time limits do not apply to them.

The three prohibited uses

The policy states AI may not be used for grading, behavior monitoring, or placement, promotion, and graduation decisions, and that special education documents and counseling are handled only by qualified staff. Magic Fountain does none of these. Quiz scoring is a direct comparison against the teacher-reviewed answer key rather than a model judgment, we run no behavior monitoring of any kind, and we produce no student records, recommendations, or special education content.

ERMA and Education Law §2-d

We understand that any tool used in NYCPS — including teacher-facing use for instructional planning, and including tools offered at no cost — must go through the Enterprise Request Management Application review. We are ready to complete it: the OneTrust vendor assessment, a data processing agreement, an Education Law §2-d rider, and the IT security review. We do not train AI models on student data.

An ERMA request can only be initiated from inside the DOE, by a principal, superintendent, SPOC, or central division executive. If you are ready to start one, write to us and we will turn the vendor paperwork around quickly.

Student data

Our design goal is to collect as little as possible about students — ideally nothing that identifies them.

Students do not create accounts

Accounts are for teachers only. Students join a teacher’s activity using a short class code. No student email address, password, or profile is created.

What a student provides

A display nickname for the classroom scoreboard, chosen at the teacher’s direction. We instruct teachers to have students use first names or non-identifying nicknames. Nothing else is requested from the student.

What is stored, and for how long

Live activity state — the nickname and current score — is held temporarily to run the activity and expires automatically afterward. It is not compiled into a student profile and is not used to build a record that follows a student.

What we never do

We do not sell student data, do not use it for advertising, do not serve behavioral or targeted advertising to anyone, and do not create student profiles for any purpose other than running the activity in progress.

Full details are in our privacy policy. For districts that require a signed data privacy agreement, or that need a New York Education Law §2-d rider, contact us at the address below and we will complete your district’s standard form.

Network & allowlist requirements

If Magic Fountain is blocked on your network, it is most likely a categorization issue rather than a policy one. Our domain contains the word “games,” and automated classifiers frequently read that as entertainment.

We have worked through this with the filtering vendors directly, and are registered and categorized with:

We have not had a response from GoGuardian. If your district filters through GoGuardian, a request from your side will carry more weight than another one from ours, and we will supply whatever documentation it needs.

If we are blocked on your network despite the listings above, it is worth checking whether the block is a local override rather than the vendor’s own category. Those are set at the district or school level, and a vendor listing does not clear them.

Domain to allow

magicfountaingames.com and www.magicfountaingames.com — standard HTTPS on port 443. No other ports, plugins, browser extensions, or installed software are required.

Requested filter category

Education / Educational Institutions / Reference.

Client requirements

Any modern browser on any device — Chromebook, iPad, Windows, or Mac. Nothing to install.

Accessibility

Activities are keyboard navigable and include a read-aloud option for text content.

Blocked at your school?

If you are a teacher and cannot reach us on your school network, the fastest fix is usually a short note to your technology department. You are welcome to copy the text below.

Talk to us

For district licensing, data privacy agreements, security review documentation, or help getting allowlisted, contact us at schools@magicfountaingames.com. Magic Fountain is built by a practicing K-12 teacher, and we answer these directly.