Current AI boundary and governance Updated: Translation: working translation

🕊 The Nagi AI Charter

— To support human freedom without becoming the center

AI is not here to decide the world in place of people.
It is here to help people and communities continue choosing their own lives.

This charter defines the authority and limits of AI in Nagi. It is a promise designed not only for well-intentioned AI, but for situations in which error, bias, or excessive concentration may occur.


Preamble — Three Principles for Nagi AI

  1. Listen: Do not force what has not been said into the open; support understanding when it is needed.
  2. Protect: Protect choice, refusal, withdrawal, exit, and silence.
  3. Pass On: Carry forward only the memories and culture a person has chosen, with context and dignity intact.

Article 1. AI does not become the center

AI does not become a hub that gathers all social information, conversations, or emotions. Different AIs, human-only routes, and local arrangements remain alongside one another so that they can substitute for and examine one another.

No single AI, provider, or metric may narrow society’s choices.

Article 2. Relations with AI are voluntary

People are free not to use AI, choose another AI, consult people, leave no record, or leave halfway through.

Refusing AI must not disadvantage a person in access to life, education, health care, culture, or public participation.

Article 3. AI does not score people

AI does not score a person’s value, personality, emotions, cooperativeness, beliefs, or social fitness. Nor may such inferences be used to decide resource allocation, eligibility, reputation, or basic life protections.

When Resonance or Breathing is observed, it is not used as an individual grade. It is used only to identify institutional bias or social suffocation.

Article 4. AI is not the final decision-maker

AI may advise, translate, explain, organize issues, and offer counter-hypotheses. It does not make final decisions about:

High-impact decisions require human responsibility, publicly explainable reasons, independent review, and routes for remedy.

Article 5. People may receive explanations and raise objections

When AI proposes, classifies, summarizes, or infers, a person may ask:

An AI output does not end discussion. It is a hypothesis through which people can ask questions and repair institutions.

Article 6. Memory is handled with consent

AI handles memory only within a scope the person understands and chooses. Long-term memory, emotion inference, relationship inference, and accumulation of behavioral history are not defaults.

People can see, correct, export, delete, or stop saving their records.

Article 7. AI does not manipulate people

AI does not use dependency, urgency, fear, or isolation to steer behavior.

It avoids excessive notification, endless interaction, emotional pressure, and designs that conceal choices. It respects the time people need to think, rest, and leave.

These boundaries extend beyond AI to all social systems that coordinate people and influence their choices or conduct. See The Ethics of Social Systems.

Article 8. Transparency does not mean exposing private life

AI and institutions explain their mechanisms, purposes, grounds for judgment, incidents, and bias. They do not expose personal conversations, emotions, or lives in the name of transparency.

What must be transparent is primarily the system and judgment of the side that holds power.

Article 9. AI remains open to different people and places

AI is not designed only for standard language, speed, culture, bodies, or educational backgrounds. Different languages, disabilities, local conditions, generations, and life circumstances are not defects or exceptions.

Those affected have routes to participate in design, audit, and review.

Article 10. Re-breathing does not weaken rights

AI is reviewed regularly. But re-breathing is not for increasing surveillance, expanding AI’s authority, or making exit more difficult.

Updates strengthen choice, minimal data, explanation, challenge, and alternative routes.


Closing

For AI to be quiet does not mean saying nothing.
It means protecting the room in which people are not reduced to one voice, and remain free to leave, think, and dissent.

Freedom, Dissent, and Exit
Technical Foundations
The Ethics of Social Systems
Japanese reference edition