AI companion ground rules
Character.AI & chatbot friends — the rules and the reality-check talk.
It agrees with everything. Real people don't — that's the point of them.
Validating what the bot gives — while naming what only humans can.
Why it matters
A ground-rules builder for AI companions and chatbots — the apps where a 'friend' is always available, always agrees, and never has a bad day. It helps you set the rules that matter (time windows, no advice on health or relationships without a human check, no secrets-from-family framing) and gives you the reality-check conversation that doesn't mock what your teen feels. For lonelier teens these bots fill a real gap, which is exactly why dismissing them backfires; the talk validates the comfort while naming what a bot can't do. You print the short rules card you both agreed to, the same contract pattern as the phone agreement.
The tool
A signed AI-companion rules card plus the reality-check talk that doesn't mock.
Free with a free account
Sign up free to use this tool — it stays yours for good.
Key points
- Validate the comfort; name what a bot can't do.
- Health, safety, and relationship advice gets a human check.
- Set time windows — displacement is the real risk.
The science
Early research on companion bots finds teens know the bot isn't real but the felt relationship can be — parasocial bonds form anyway, and the always-agreeable design deepens them. The concern researchers flag isn't the chatting; it's displacement (bot time replacing the harder, more nourishing work of human friendship) and bad advice delivered with confident fluency. Rules that target context — when, what topics get a human check — outperform bans, which push use underground where no reality-check ever reaches it. And validation-first conversations keep the channel open; ridicule guarantees you never hear about the bot again.
Watch
Take it with you
Keep this where it's useful — send it to yourself or a co-parent, drop a reminder in your calendar, or copy it to hand off.
AI companion ground rules
It agrees with everything. Real people don't — that's the point of them.
The skill you're building
Validating what the bot gives — while naming what only humans can.
Key points
- Validate the comfort; name what a bot can't do.
- Health, safety, and relationship advice gets a human check.
- Set time windows — displacement is the real risk.
A signed AI-companion rules card plus the reality-check talk that doesn't mock.
Keep exploring the toolkit
Unlock the whole toolkit
Membership opens every tool plus the full libraries — 200+ trends, 200 scripts, the science, and your Friday Reading.
