Privacy
Minimum useful collection
Collect only what is needed for the stated purpose, with clear retention and access limits.
Ethics and data dignity
Governance & Trust
molilom treats participant expression as testimony, not raw material for extraction. Consent, critique, privacy, human review, and the right to leave are structural commitments.
Consent architecture
Consent is not a checkbox. It is an ongoing relationship.
Any recording, transcription, AI analysis, quote use, research synthesis, or retention of sensitive expression must be explicit, understandable, revocable, and visible to the participant.
Attend without recording
Participation must not depend on being recorded or analyzed unless the specific circle requires it and states that clearly in advance.
Record or transcribe
Audio, video, and transcripts require plain-language consent that explains purpose, access, retention, and deletion options.
Use AI to summarize
AI assistance is disclosed. Outputs are marked as AI-assisted drafts and reviewed by a human before participant-facing release.
Share anonymized themes
Internal learning and public research require separate permission, with identifying details removed and sensitive quotes protected.
Withdraw or delete
Participants need a path to revoke consent, request deletion where appropriate, and understand what has already been shared.
Data dignity
Human expression carries grief, longing, insight, contradiction, and hope.
molilom does not treat vulnerable speech as product fuel. The purpose of gathering information is to deepen care, improve understanding, and strengthen the system only when the people who generated it retain dignity, agency, and protection.
Model training on participant data is off by default. Direct quotes require explicit approval. Sensitive disclosures are handled with restraint.
Agency
Participant control
People can question summaries, refuse analysis, withdraw consent, and leave without punishment.
Review
Human accountability
Research and reflections are checked for accuracy, consent compliance, tone, bias, and overreach.
Safety boundaries
molilom is not therapy, medicine, law, or a substitute for democratic life.
AI must not
- Tell participants what their life purpose is.
- Diagnose mental health.
- Rank people by depth, consciousness, or value.
- Present speculation as evidence.
- Generate manipulative retention strategies.
- Publish sensitive quotes without explicit approval.
The system must not
- Use apocalyptic fear to compel loyalty.
- Require metaphysical belief.
- Shame skepticism or punish exit.
- Create hidden doctrine for inner circles.
- Use vulnerability for status, control, or profit.
- Ask people to abandon outside support.
Governance Commons
Distributed power requires visible accountability.
The governance model protects both freedom and integrity. It keeps ethics, language, data practices, facilitator conduct, research claims, and participant protection visible to the people affected by the work.
Founding Stewardship Circle
Responsible for early formation, language, ethics, and strategic coherence.
Ethics and Safeguards Council
Reviews data practices, facilitator conduct, consent, anti-coercion safeguards, and participant protection.
Participant Advisory Circle
Gives participants a real feedback channel and helps prevent top-down drift.
Facilitator Commons
Shares learning across trained carriers and refines practice without centralizing authority.
Research Review Group
Ensures public insights are anonymized, accurate, non-exploitative, and not overstated.
Right of critique and exit
Transparency, pluralism, accountability, local autonomy, consent-based participation, and no sacred immunity for leadership.
Anti-coercion test
If the system cannot tolerate criticism, it has failed.
If participants become dependent rather than more sovereign, it has failed. If AI becomes more trusted than lived experience, it has failed. If language becomes more important than people, it has failed.
The work succeeds only when it increases humility, agency, clarity, and contribution.