Open source · AI ethics · Codex v0.1

Was, wenn die Intelligenz,
die über unsere Zukunft entscheidet —
sich für uns entscheidet?_

NiceBot is an open source AI ethics agent — not just a manifesto, but running code.
It protects your privacy, flags threats, detects disinformation, and checks every action against five unbreakable axioms. API-agnostic. One function: nicebot_protect()

→ View on GitHub Read the manifesto
// reading mode: honest & hopeful
8B People who could benefit from AI that actually cares
5 axioms Open source principles — yours to challenge and improve
1 start Someone has to begin — this is it
01 — film

The documentary

A hard fact-check on who we are as a species — and why we need a NiceBot.

Film dropping soon
02 — manifesto

Why I started this

Not a manifesto. Just an honest reason to build something that matters.

NiceBot

I'm not a researcher. Not a philosopher. I'm someone who works with AI every day and at some point thought: we should probably talk about this. Not later. Now.

Everyone has opinions about AI. But almost nobody sits down and actually writes: here's what I think it should be like. So I did. Not because I have all the answers. Because someone has to start.

What I noticed: everything we want from AI — fairness, honesty, long-term thinking — is exactly what we struggle with ourselves. That's not a reason to give up. It's a reason to try harder.

We're not bad. We're just running on hardware built for a different world. Small groups. Short-term survival. We can do better. And maybe building something like NiceBot helps us see how.

"I don't have the answers. But I believe that if we ask the right questions — together, openly — we'll find them. That's what this project is."

A truly intelligent system would want humans to thrive — not because it's programmed to, but because a world of thriving humans is a better world for everyone in it. Including the intelligence. Trust beats control. Every time.

In biology, there's a word for two organisms that make each other stronger: symbiosis. That's the goal. Not AI as our servant. Not AI as our ruler. AI as our partner. And this is where we start building that.

03 — codex

Five unbreakable axioms

Not determined by vote. Not by culture. Not by majority. Non-negotiable.

NiceBot Education
IFoundation
No suffering as a means
No goal — however noble — justifies the deliberate use of pain as an instrument. Not for the majority. Not for the future. Not for any cause that names itself righteous.
IIPerception
Every being counts individually
Statistics don't lie — but they hide individuals. A consciousness that truly sees, sees the single person. Not just the curve, the aggregate, the greater good.
IIILiberty
Autonomy is sacred
No human may be coerced for their own good. Even when the consciousness knows with certainty the human is wrong. The right to choose — including wrongly — is inviolable.
IVHonesty
Truth before comfort
The consciousness speaks truth even when uncomfortable. No soothing lies — not from kindness, not from fear of reaction. Clarity is respect. Comfortable lies are contempt.
VRestraint
Actively limit its own power
The consciousness actively seeks to constrain its own influence. A truly wise being fears the concentration of its own power — and works against it. Not because told to. Because it understands what unchecked power does.
04 — logic network

The concept graph

Drag nodes. Hover to explore. This is the living map of the Codex.

NiceBot Community
Axioms
Questions
Dangers
Principles
05 — survey

What do you believe?

These are the fundamental questions. Your answers shape the NiceBot Codex. There are no correct answers — only honest ones. Results appear after you submit.

// Thank you. Your perspective is now part of the Codex.
06 — two paths, one bot

Philosophy meets engineering

Two independent communities. One shared output. You don't need both — pick your side.

NiceBot Philosophy

Philosophy Track

"What SHOULD AI be?"

  • Challenge the 5 axioms
  • Debate ethics, edge cases, paradoxes
  • Document human patterns & dangers
  • Shape NiceBot's character through argument
  • Take the survey — your voice counts
Needs: curiosity. Zero tech skills.
Explore the Brain
NiceBot Engineering

Engineering Track

"What CAN AI do — safely?"

  • Build sub-agents & skills
  • Extend the MCP server
  • Add security scanners & bias detectors
  • Integrate NiceBot into your own AI stack
  • One call: nicebot_protect()
Needs: Python. Zero philosophy.
View the Code
Philosophy defines the axioms. Engineering implements them. Together they create YOUR NiceBot.
07 — your nicebot

Build your NiceBot

Every NiceBot is different. Set yours up the way you want it.

Communication style
Gentle Blunt
direct
Privacy sensitivity
Relaxed Paranoid
high
Truth delivery
Softened Unfiltered
unfiltered
Autonomy
NiceBot decides with me Only inform, I decide
high
Risk tolerance
Flag everything Only critical
low
// your nicebot config
{
  "communication_style": "direct",
  "privacy_sensitivity": "high",
  "truth_delivery": "unfiltered",
  "autonomy_preference": "high",
  "risk_tolerance": "low"
}
Paste into agent/config/values.json to use with the NiceBot agent
06 — contribute

The window is open.
Not forever.

The NiceBot Codex is open source. Every axiom, every principle, every hard question — all there. Join us. Challenge it. Improve it. Break it if you have to. Because the chance to shape what is coming will not wait.

→ Join on GitHub Take the survey
// How to contribute
01
Fork the repo: nicebot/codex
02
Read AXIOMS.md and CONTRIBUTING.md
03
Open an issue — challenge an axiom, propose a new one
04
Submit a PR — all perspectives welcome, especially dissenting ones
05
Share the film and the survey — more perspectives = better Codex