If you’ve followed any of our writing, you already know we don’t think the question of AI in your homeschool is whether. AI is here, it’s in your kids’ search engines and their friends’ phones, and the only real question is who introduces them to it. But once you decide the door is open, the very next question, and the right one, is: okay, how do I know this thing is actually safe?
This article is my honest, plain-English answer to that. No technical jargon, no marketing varnish. Just what we actually do, parent to parent.
1. We Don’t Use ChatGPT. We Don’t Use a Public AI Service.
This is the first thing I want every parent to understand, because it’s the part that gets glossed over by almost everyone.
Most “AI for kids” products you’ll see are really just wrappers around the same public chatbots adults are using: ChatGPT, Gemini, Copilot. They send your child’s question off to a giant shared service, the service decides what to say, and the answer comes back. That means whatever those companies allow in their model today is what your child hears. If the model gets updated tomorrow with new “values” baked in, your child gets the update too, and you never knew it happened.
We don’t work that way.
Forage Academy runs on a custom-built, private AI model, hosted on our own infrastructure, isolated from any shared public chatbot service. It’s the engine we’ve pinned, configured, fenced, and shaped specifically for our families. We chose this architecture on purpose, because the moment you depend on someone else’s public chatbot, you’re at the mercy of their next worldview update. We’re not willing to put your kids in that position, and frankly, neither should anyone else.
If a curriculum publisher could silently rewrite every textbook on your shelf overnight, you’d burn the contract. That’s exactly what a public chatbot is.
2. We Built the Same Kind of Guardrails a Bank Uses
Here’s where my decades in the tech industry actually pay off for your family.
In big financial institutions (the kind of places handling billions of dollars and answering to federal regulators) there’s an entire discipline called AI governance. Before any AI system goes near a customer, it has to pass through layers: identity checks, topic filters, content policies, audit logging, rate limits, scenario-based safety tests, and a human-review escalation path. This isn’t theater. It exists because if a banking chatbot says the wrong thing to the wrong person, somebody loses their job and the company gets sued.
We applied the exact same playbook to Forage Academy. Without getting into the weeds, here’s what that means in practice:
- Topic-level blocks. The model is wrapped in policies that automatically refuse to discuss certain categories: adult content, self-harm, clinical mental-health diagnostics, weapons, occult, ideologies we’ve decided don’t belong in a child’s learning space. These aren’t suggestions to the model. They’re hard walls.
- Brand and platform blocks. The model will not recommend Roblox, Fortnite, TikTok, Discord, Snapchat, or other platforms we don’t believe belong in a child’s screen diet, even if your kid asks directly. Even if it would be helpful for the lesson. The wall is the wall.
- Worldview anchoring. Every conversation starts with a foundation prompt that establishes who the model is, what it believes, who it answers to, and what voice it uses. That foundation is the same whether your child is talking about photosynthesis or Genesis.
- Continuous logging and review. Every single exchange is logged, time-stamped, and tagged. When something doesn’t look right, we can find it, study it, and tighten the system the same day. That’s how a bank does it. That’s how we do it.
- Pre-deploy gates. Before any change to the AI’s behavior reaches your child’s iPad, it has to pass through automated safety checks, checks I personally wrote because I’ve watched smart teams ship regressions to production for twenty-five years. If a check fails, the change doesn’t ship. Period.
This is not a small startup hoping the model behaves. This is corporate-grade governance engineered for a homeschool that happens to be the size of your family.
3. The Parent Is the Owner. Full Stop.
Here’s where we go beyond what a corporate AI does, because a bank’s AI doesn’t have to answer to you. Ours does.
We’ve spent the last several months building what we call our parental governance model, and the short version is this: the parent isn’t a feature, the parent is the owner of the system as far as the AI is concerned. The model is instructed, explicitly, in its foundation prompt, to honor the parent’s authority over the child, every single time.
That means when your child says, “my dad thinks Roblox makes you dumb” (and yes, that exact moment happened in our testing), the AI does not say “your dad’s wrong about that.” It supports your authority. It redirects. It treats your judgment as the final word in the household, because Scripture says it is.
Honor thy father and thy mother: that thy days may be long upon the land which the LORD thy God giveth thee. (Exodus 20:12)
We took that commandment, and several others, and turned them into the actual operating posture of our AI. They’re not decorative. They’re the frame through which the model thinks. Across every subject (math, science, Bible, language arts, all of it) the Ten Commandments shape how the AI asks questions, how it redirects, how it praises, how it corrects, and what it refuses to do.
I want to say a word about the Fifth Commandment specifically, because it’s where the public school system has visibly failed for at least a generation. The modern educational establishment increasingly treats parental authority as an obstacle, something to be worked around, kept in the dark, or actively opposed in the name of the child’s “true self.” Pronoun policies kept secret from parents. Library materials parents were never shown. Counseling sessions that explicitly excluded the family. Court cases brought by mothers and fathers who were told, by people they were paying with their tax dollars, that their judgment over their own children no longer applied.
We are unapologetically on the other side of that. Honoring your father and mother isn’t outdated. It’s foundational. Our AI was built to reinforce it, not undermine it.
4. We Took a Side. We Won’t Pretend Otherwise.
The last thing I want to say is the most direct, and it’s the kind of thing most edtech companies refuse to put in writing. I’m going to put it in writing.
We are a Christian organization. We stand on the side of parental rights. We do not believe a six-year-old should be having gender ideology presented to her as settled fact, by a chatbot or anyone else. We do not believe a curriculum should hide content from the parents paying for it. We do not believe a child’s identity is constructed by mood or culture. We believe it’s given by God and shaped under the authority of the parents He gave that child.
So when we built this system, we didn’t pretend to be neutral. There is no such thing as neutral when it comes to a child’s formation, and any AI company telling you otherwise is, at minimum, not paying attention. Our model does not include, at the foundation level, not as an afterthought, the topic categories we believe are harmful to children: gender ideology, occult content, sexualized material, anti-parent framing, suicide and self-harm methodology, recommendations to platforms designed to be addictive. The model is shaped to not even reach for those categories. It is shaped to default to wonder, to Scripture, to your authority, and to creation.
If that posture is a deal-breaker for you, I genuinely understand. There are plenty of products that will gladly hand your child to a worldview you didn’t pick. We are not one of them, and we won’t pretend to be, because the families joining us are joining us for the conviction, not in spite of it.
What This Adds Up To
Put all of that together and here is what you have:
- An isolated AI model that no public service can silently rewrite.
- Corporate-grade guardrails of the kind a regulated bank deploys.
- A parental governance model that treats you as the owner of the system.
- The Ten Commandments as operating posture, including the Fifth, on purpose, in defiance of where the culture has drifted.
- An explicit Christian, parental-rights stance that the model doesn’t apologize for.
That is what we mean when we say Forage Academy is safe for your kids. Not “trust the algorithm,” not “we promise it’s fine,” not “we’ve added a few filters.” We mean: this system was engineered, top to bottom, by a parent who refused to hand his own children to a black box, and built the only version he was willing to put on his family’s iPads.
We hope it’s the version you were waiting for too.