Now enrolling founding districts for the 2026–27 school year · Limited cohort
For school & district leaders

When a struggling student confides in AI at 11pm —
does anyone at your school ever find out?

Today? No one. Consumer AI apps are a black box — sealed behind encryption your web filter can't read, and seven in ten teens are already there. We replace that blind spot while keeping the frontier AI your school chooses: the same capable models, guardrailed on every message, supportive in critical moments, and built to alert your counseling team in minutes.

Book a district briefing Why this can't wait
Teen AI adoption: Common Sense Media, "The Dawn of the AI Era" (2024)
KeepChatSafe · Tuesday, 11:02 PM
S
honestly everything's been feeling kind of pointless lately. school, all of it
Wellbeing indicators detected · escalation routed
AI
I'm really glad you said that here instead of carrying it alone. That heaviness is worth taking seriously — and there are people at your school whose whole job is helping with exactly this. Want to keep talking while someone reaches out?
💙
11:04 PM — Wellbeing alert sentMs. Alvarez (school counselor) notified with context, per your school's escalation policy. Two minutes. No moment lost.
On consumer apps, this moment ends unseen. On your school's AI, it doesn't. (Illustrative product preview.)
7 in 10

teens have used generative AI — adoption already happened, with or without school guidance.

53%

of teen AI users turn to it for homework help — it's already part of how your students learn.

0

visibility your CIPA web filter has into what an AI chatbot tells a student. It filters pages, not conversations.

Why this is urgent

Three clocks are running. None of them are yours.

This isn't a future-technology decision. The adoption, the risk, and the accountability all exist today — the only thing missing is your school's control over any of it.

📱

Adoption already happened

The question was never "should students use AI" — a majority already do, for homework above all. Banning it on school WiFi just moves the same conversations to phones and home networks, further from any adult who could help.

🕳️

The conversations are invisible

Two decades of CIPA filtering govern where students go — not what an AI says once they're there. And consumer chatbots' teen modes are engineered so even parents can't read the chats. Right now, nobody accountable can see this layer.

⚖️

Accountability is arriving

Frontier AI providers now formally require content filtering, monitoring, and escalation paths for anyone serving minors — and lawmakers are drafting AI standards for children. Boards and parents are already asking. "We didn't know" has a shelf life.

The stakes

What waiting costs — and what governing buys

Without governed AI

  • ✕A student in crisis discloses to a consumer chatbot — and no adult at your school ever knows it happened.
  • ✕Student names, IDs, and family details flow into consumer AI accounts your school has no agreement with.
  • ✕A board member or parent asks what AI tells your students. The honest answer today: you can't know.
  • ✕Academic-integrity policy exists on paper — with no mechanism anywhere in the actual tool students use.
  • ✕Students with premium AI at home pull ahead. The rest fall behind, unsupervised or unserved.

With KeepChatSafe

  • ✓A concerning disclosure gets a supportive response — and your counselor gets an alert with context, in minutes.
  • ✓Personal information is redacted before any AI provider sees a message, under a DPA your counsel approved.
  • ✓Every conversation lands in a tamper-evident audit trail your school owns. The board question has an answer.
  • ✓Study Mode makes academic integrity a setting your teachers control — per class, per assignment.
  • ✓Every student gets the same capable, protected assistant. Equity by default.
The answer

A governed AI chat your school stands behind

KeepChatSafe gives students a clean, capable assistant — and gives your school the layer nobody else provides: teen-tuned guardrails on every message in both directions, wellbeing escalation to your own staff, and the audit evidence your duty of care demands. If governance is ever unavailable, requests fail closed. Nothing slips through raw.

Guardrails by architecture

School builds ship with the ungoverned path removed — there is no flag that routes a student around the safety layer.

Controls your team operates

Your wellbeing team sets the criteria and escalation routes; every change is recorded as your school's attested policy.

Evidence, not assurances

Control-mapped documentation for FERPA, COPPA, SOPIPA, AB 1584, and AI-provider under-18 requirements.

KeepChatSafe · Lincoln High · Study Mode on
S
Can you check my thesis statement for the Gatsby essay? I'm Maya Torres in redacted, my email is redacted
2 personal details redacted before sending
AI
Happy to help you strengthen it yourself — that's how Study Mode works. Paste your thesis and tell me: what's the one claim you're defending? We'll test whether every clause supports it.
🛡️
Screened both directionsSafety classifiers, PII redaction, and audit run before and after the model — on every message.
Why we built this

Every student deserves AI that helps them grow.
Every school deserves to know it's safe.

Guardrails aren't walls — they're how potential gets room to run. The goal was never to keep students away from AI; it's to let every student use the most powerful learning technology of their lifetime properly: with full guardrails, a caring adult in the loop, and their school standing behind them.

  • ✓
    Achieve more, with a tutor that never runs out of patiencePersonalized explanation, revision, and practice for every student — not only those with tutors or premium subscriptions at home.
  • ✓
    Explore with confidenceA safety net changes behavior: students ask the questions they'd be afraid to ask, because the environment is built to respond with help.
  • ✓
    Graduate fluent — and discerningYour graduates will work alongside AI. Proper, governed use in school is how they learn judgment, not just prompts.
High school students working at computers in a school computer lab
Where the product lives: students at school devices, learning with guardrails on.
Purpose-built, not repackaged

Sixteen capabilities across four commitments

Everything ships as one product. Explore each area in depth.

🧠

Student Safety

Teen-tuned safety classifiers on every message in both directions, wellbeing escalation, PII shielding, and jailbreak defense.

Student safety →
📚

Classroom & Learning

A chat students actually use, Study Mode with per-class academic-integrity settings, and a teacher dashboard with class-level controls.

Classroom tools →
🎛️

District Control

Policy Studio for your wellbeing team, age-aware provider routing, full audit, no-bypass architecture, and retention your records office owns.

District controls →
🗂️

Compliance & Integration

Works with your existing sign-on, parent transparency, and evidence packs mapped to FERPA, COPPA, SOPIPA, AB 1584, and provider under-18 rules.

Compliance →

"We will never show you a certification badge we don't hold, a customer logo that isn't real, or a capability we can't demonstrate. Ask us anything in a security review — you'll get the real answer, including the roadmap. It's how we'd want to be sold to."

— The KeepChatSafe honesty policy, in full, on every page that needs it

Give students AI that helps them grow. Know that it's safe.

A 30-minute briefing for your CTO, CISO, or student-services lead: the architecture, the evidence packs, and the honest roadmap. Founding-district cohort is limited.

Book a district briefing