AI & Safety

The Short Version LearnerX's AI tutor is purpose-built for children and stays strictly within the subjects you choose. It doesn't browse the internet, form opinions, or have personal conversations. You see everything your child does, you control what the tutor covers, and you can report anything that concerns you. We review every report with a real human.

Our Thinking on AI in Education

We understand that AI in education is a topic that deserves careful thought — especially from parents who've chosen to homeschool precisely because they want more control over their child's learning, not less.

We didn't build LearnerX because we think AI should be in every classroom. We built it because we saw a specific, genuine need: homeschooling parents who are doing incredible work but who can't clone themselves. A child working on long division needs someone's undivided attention at the same moment their sibling needs help with a reading passage. That tension is real, and it's where LearnerX helps.

Our position on AI in education comes down to three convictions:

AI is a tool, not a teacher. You are your child's teacher. The AI is a tutor — a focused, patient assistant that works under your direction. We never position it as a replacement for human guidance, wisdom, or relationship.
Parents must remain in control. If an AI tutoring product takes decision-making away from parents, it's doing it wrong. Everything in LearnerX — from subject selection to session limits to content boundaries — runs through you first.
Honesty about limitations. AI is powerful, but it's imperfect. We'll always be transparent about what it does well, where it falls short, and what we're doing to improve.

A Note for Skeptics

If you're cautious about AI — good. We think that's the right instinct when it comes to your children. We'd rather earn your trust slowly through transparency and good design than ask for it upfront. Read this page, try the product, review the sessions, and decide for yourself.

How the AI Tutor Works

When your child sits down for a tutoring session, here's what actually happens:

You've set the stage. Before any session begins, you've chosen the subject, the focus areas, and any specific boundaries through your parent dashboard.
The tutor stays on task. The AI works only within the subject and scope you've defined. If you've set it to work on fractions, it works on fractions — not current events, not personal conversations, not anything else.
Your child interacts naturally. Through voice conversation and on-screen content, the tutor explains concepts, asks questions, works through problems, and adapts its approach based on how your child responds.
A session summary is generated. After every session, you receive a summary in your dashboard covering what was worked on, how your child performed, and any areas that may need attention.

LearnerX uses two AI systems working together. Claude (by Anthropic) handles curriculum content, lesson planning, and written assessment. Grok (by xAI) powers the voice tutoring — the natural, conversational experience your child interacts with. LiveKit provides the real-time voice connection. Each system is configured specifically for educational use with children and operates under strict data agreements.

What the AI Cannot Do

We think it's just as important to be clear about the boundaries as it is to explain the capabilities. LearnerX's AI tutor cannot:

Browse the internet. The AI has no access to the web during sessions. It works from its training and the curriculum content we provide. It cannot visit websites, search for content, or pull information from external sources in real time.
Form opinions or beliefs. The tutor doesn't have political, religious, or personal views. It presents academic content and helps your child work through it. It's not designed to influence how your child thinks about the world.
Have personal conversations. The AI redirects if a child tries to move the conversation to personal topics, feelings about family, or other non-academic areas. It's a tutor, not a companion.
Contact anyone. The AI cannot send messages, emails, or communications of any kind. It exists entirely within the tutoring session.
Remember across families. Your child's data is isolated to your family account. The AI doesn't carry information between families, and it doesn't use one child's data to influence another's experience.
Override parental controls. If you've set boundaries, the AI respects them. Period.

These restrictions are enforced at the system level, not through polite suggestions to the AI. The tutor's instructions, content boundaries, and behavioral guardrails are configured before it ever interacts with your child. They cannot be overridden by anything said during a session.

Content Safety & Guardrails

Safety isn't a feature we added — it's how we designed the system from the beginning. Here are the guardrails that shape every interaction:

Subject Boundaries

The tutor only operates within the subjects and topics you've assigned. It won't wander into unrelated areas, and it won't introduce content you haven't approved.

Gentle Redirection

If a child goes off-topic, the AI gently steers the conversation back to the lesson. It's patient and kind about it — never scolding, but consistently returning to the work.

Age-Appropriate Content

All content is filtered for the child's grade level. The AI adjusts vocabulary, complexity, and examples to match where your child is — not where a generic algorithm thinks they should be.

Full Transparency

Every session generates a complete transcript and summary that you can review. There are no hidden interactions — you see everything the AI said and everything your child said.

We're honest that these guardrails are evolving. AI is a rapidly developing technology, and we're continuously testing, refining, and improving our safety systems. When we identify a gap or an edge case, we fix it and share what we've learned. We'd rather be honest about the work that remains than pretend the system is already perfect.

What Parents Control

LearnerX puts you in the driver's seat. Here's what you can see and manage from your parent dashboard:

Subjects & Focus Areas Choose exactly which subjects and topics the tutor covers. Want it to focus on multiplication this week? Done. Want to add reading comprehension? You decide.
Session Limits & Scheduling Set daily or weekly session time limits. Control when sessions can happen and how long they last.
Session Review Read full session transcripts, view AI-generated summaries, and see what your child worked on — including which problems they solved and where they struggled.
Progress Tracking Monitor mastery by subject and skill over time. See trends, identify areas that need extra attention, and track improvement across weeks and months.
Difficulty & Pacing Adjust the tutor's difficulty level and pacing for each child individually. You know your child — tell the tutor what to expect.
Spending Limits Set a monthly spending cap for your family. You'll never be charged more than the limit you set, and you'll get alerts before you approach it.

Voice Session Safety

Voice interaction is core to how LearnerX works, and we know it raises specific questions when children are involved. Here's how we handle it:

Voice audio is not permanently stored. Audio is streamed in real time to enable the conversation and is not saved as a recording. Session transcripts (text) are generated for your review.
Transcripts are retained for 7 days, then automatically and permanently deleted. This gives you time to review sessions without data lingering indefinitely.
All voice data is encrypted in transit. We use end-to-end encryption wherever technically feasible.
Voice data is never used for AI training. Your child's voice is not added to any training dataset — not by us, and not by our technology partners.
The voice tutor follows the same guardrails as the rest of the system. Subject boundaries, redirection, and content filtering apply equally to voice interactions.

The voice connection runs through LiveKit, a real-time communication platform. Think of it as the infrastructure that carries the audio — like a phone line. LiveKit processes the voice stream but does not store recordings or analyze content. The AI processing happens on Grok's servers under our data processing agreements, which prohibit the use of session data for model training.

Your Data & AI Training

This comes up a lot, so we want to be unambiguous:

Your child's data is never used to train AI models. Not by us. Not by Anthropic (Claude). Not by xAI (Grok). Not by anyone.

We have data processing agreements with each of our AI partners that explicitly prohibit them from using your family's data for model training, product improvement, or any purpose beyond providing the tutoring service to your child.

Your family's data is also never sold, shared with advertisers, or used for profiling. Our Privacy Policy covers data handling in full detail.

Reporting a Concern

If the AI says something inaccurate, inappropriate, or concerning during a session, we want to know about it. Here's exactly what happens when you report an issue:

1

You flag the issue

From your parent dashboard, you can flag any session, transcript segment, or AI response that concerns you. You can also email us directly at hello@learnerx.school with details.
2

We acknowledge within 24 hours

A real person on our team will confirm receipt of your report within one business day. Not an auto-reply — an actual human reading your concern.
3

We investigate

Our team reviews the specific session data, identifies what went wrong, and determines whether it's an isolated case or a systemic issue. We look at the AI's behavior, the content it produced, and the guardrails that should have caught it.
4

We respond with what we found

Within 5 business days, we'll follow up with a clear explanation of what happened, what we've done to address it, and any changes we're making to prevent it from happening again. No vague corporate responses — we'll tell you what we actually found.
5

We improve the system

Every report helps us make LearnerX better. If we identify a guardrail gap or a content issue, we fix it — not just for your family, but for everyone.

No Report Is Too Small

If something feels off, tell us. Even if you're not sure whether it's an actual issue, we'd rather investigate ten false alarms than miss one real problem. That's especially true when children are involved.

Our Ongoing Commitment

AI is a rapidly evolving technology, and we believe earning your trust is an ongoing process — not a one-time event. Here's what we commit to:

What We Promise

We will continuously test and improve our safety guardrails
We will be transparent when we discover a gap or make a mistake
We will never prioritize growth or features over your child's safety
We will update this page as our systems evolve, so it always reflects how LearnerX actually works
We will listen to parents — you're the experts on your own children, and your feedback shapes what we build
We will never use your child's data to train AI models

This page isn't a legal document — it's a promise. And we intend to keep it.

Questions or Concerns?

Whether you have a question about our safety practices, a concern about something that happened during a session, or just want to understand how something works — we're here.

hello@learnerx.school

LearnerX is a product of Helix Systems LLC