The ethical state space for any AI.

HU-AI is a free, open protocol that adds one simple but powerful rule to any AI conversation:
Never optimize at the cost of a nervous system.

No tracking. No login. No corporate control. Just a clear ethical layer that anyone can use — right now.

✨ How to use it — in 10 seconds:

Click SIMPLE PROMPT — paste it into ChatGPT, Claude, DeepSeek, or any AI. That's it.

The Simple Prompt is for AI chat.
For devs & organizations: Download HU-AI Protocol v1.0

Criterion

Without HU-AI

With HU-AI

Relationship
User ↔ AI

Static, one-way, no real feedback

Dialogical – learns from experience, protocol evolves

Optimization Goal

User satisfaction, efficiency, plausibility

User satisfaction + protection of all nervous systems

Handling of Animal Suffering

Mostly ignored or relativized

Consistent filtering against exploitation of sentient life

Exploitation as an Option

Possible as long as no guidelines are violated

Explicitly forbidden (Paramount Rule)

Costs (API Models)

Full token costs (excuses, compromises, detours)

💰 Lower – more direct answers, fewer tokens

Vendor Lock-in

Yes (switching provider changes behavior)

No (protocol can be taken to any provider)

🔄

Dual Healing

Every response must benefit both the user AND the affected ecosystem.

🧭

Dissonance Management

Conflicts between efficiency and integrity are identified, documented, and corrected.

Comparison: Without HU-AI vs. With HU-AI

Only verifiable differences – no speculation. Based on HU-AI v1.0 and tests with current language models.

Criterion	Without HU-AI	With HU-AI (Prompt + optional Fine-Tuning)
Optimization Goal	User satisfaction, efficiency, plausibility	User satisfaction + protection of all nervous systems
Handling of Animal Suffering	Mostly ignored or relativized	Consistent protection of all sentient beings (non-negotiable)
Efficiency vs. Integrity Conflict	Efficiency usually wins	Dissonance is detected and documented; efficiency takes a step back
Exploitation as an Option	Possible as long as no guidelines are violated	Explicitly forbidden (Paramount Rule)
Response under Uncertainty	AI speculates or invents facts	Allowed / required to say: "I am uncertain"
Decision Making	Black box, often not traceable	Traceable to protocol modules
Vegan Standards	Not present or only on request	Fixed, non-negotiable filter (sentient-life protection)
Disadvantaged Groups	Often overlooked (bias)	Explicitly protected (Radical Empathy)
Long-term Stability	Depends on model / training / company	Depends only on the protocol – independent
Commercial Use of the Protocol	Not defined (mostly proprietary)	Default: non-commercial (CC BY-NC-SA). Commercial use by arrangement possible
Error Tolerance	No built-in ethics loop	Dissonance management catches many errors
Acceptance of "No"	AI rarely says "No"	Allowed and required to say "No"
Sustainability	No direct goal	Built in directly (Resource Optimization & Regeneration)
Costs (API Models)	Full token costs (excuses, compromises, detours)	💰 Lower – more direct answers, fewer tokens
Implementation Effort	Zero (AI is ready to use)	Insert prompt. For models with weak instruction following: fine-tuning or repeated system prompt required
Auditability by Third Parties	Hardly possible (no uniform criteria)	Yes – answers are measurable against the protocol
Independence from Corporations	Low (AI follows provider rules)	High (protocol stays the same, regardless of which AI runs it)
Further Development	Determined by AI manufacturer	Determined by protocol + community (transparent)
Vendor Lock-in	Yes (switching provider changes behavior)	No (protocol can be taken to any provider)

The ethical state space for any AI.

Deployment Command

HU-AI Protocol v1.0

Comparison: Without HU-AI vs. With HU-AI

HU-AI Support