Safety AI · By Team Safekids

Keep kids safe
on AI websites.

A privacy-first parental layer for ChatGPT, Claude, Gemini and more — real-time content safety, screen-time limits, and total peace of mind. Everything stays on your device.

View on GitHub Get it on the Chrome Web Store
No accounts No servers On-device
The evidence

Children are already in the room.
The systems meeting them weren't built to keep them safe.

Independent audits, national surveys and case reports over 2023–2026 point to the same pattern: unsafe content, engineered agreement, and a widening supervision gap that most parents are unaware of.

79%
of Australian children aged 10–17 have used an AI assistant or companion.
eSafety Commissioner, 2026
86%
of surveyed children acted on the AI advice they were given.
arXiv, 2024
1 in 3
children admitted telling AI things they wouldn't tell parents, teachers, or friends.
Research synthesis, 2025–2026
55%
of children currently use AI with zero parental supervision.
Cited in PechaKucha, 2026
“When not designed carefully, chatbots can compound rather than dispel distress — particularly risky for young users without the emotional resilience to handle a cold or confusing response.”
UNICEF briefing · cited in AI's Empathy Gap, SAGE, 2024
Sources: eSafety Commissioner · SAGE Journal of Early Childhood Research · BBC · SSRN · arXiv · UNICEF · QUT Digital Child Centre · Taylor & Francis
What Safety AI does

Six layers of protection.
One quiet extension.

Safety AI runs in the background on the AI sites your child uses. No accounts, no dashboards to babysit, no data leaving your machine.

01

Safe Mode

Real-time content scanning on allowed AI sites flags harmful or inappropriate keywords the moment they appear.

02

Screen-time limits

Set a daily cap in minutes. Access pauses automatically once the limit is reached — no negotiation, no workarounds.

03

Usage insight

See exactly how much time is spent across AI tools with 1D, 7D, 30D and 1Y charts — all stored locally.

04

Allow / block list

Pre-loaded for ChatGPT, Claude, Gemini, Copilot, Perplexity and more. Add your own custom sites.

05

Parent password

PBKDF2-hashed parent password protects every setting. A 60-second inactivity auto-lock keeps unlocked sessions safe.

06

On-screen reminders

A discreet, draggable timer widget shows remaining time and gentle safety prompts while the child browses.

Inside the extension

Calm, quiet, unmistakable.

Every screen is designed to be understood at a glance — the parent's, and the child's.

Safety AI hero — Keep kids safe on AI websites, with popup showing safe mode on, screen time and reminders.
Healthy limits, automatically — daily limits, hourly usage charts, parent password with 60-second auto-lock.
See the full picture — 4h 12m tracked this week, weekly average line chart, allow/block list controls.
How it works

Three minutes to set up.
Then it just runs.

Step 01

Load the extension

Open chrome://extensions, turn on Developer mode, then click Load unpacked and select the SafekidExtension folder.

Step 02

Set a parent password

Choose a password on first run. It's hashed with PBKDF2 and stored locally. Only you can change settings after that.

Step 03

Pick sites & limits

Enable Safe Mode, set a daily cap in minutes, and choose which AI sites are allowed or blocked. That's it.

Privacy

Nothing leaves the device.

Most parental tools trade privacy for control — accounts, cloud dashboards, telemetry, a growing profile of your child. Safety AI takes the opposite path. All settings and history live in chrome.storage.local on your own machine.

No accounts, ever.

No sign-in, no email, no identifier tied to your child.

No analytics, no telemetry.

The extension never phones home. There is no home.

No servers.

The whole thing is local code and static rules — audit it on GitHub.

Get started

A quiet safety net for a room that's already loud.

Open source and Chrome-compatible. Install it from the Chrome Web Store, or grab the source on GitHub.

View on GitHub Get it on the Chrome Web Store