An open-source framework to defend vulnerable minds from semantic manipulation in the age of AI.
Developed as part of the LORI Framework.
As language models and recommendation systems grow more persuasive, semantic influence over emotions, decisions, and beliefs is no longer hypothetical — it is already shaping neurodivergent users, children, and elders disproportionately.
This repository provides modular defenses to protect:
- ADHD / ASD / HSP users (Neurodivergent)
- Children and teens exposed to emotionally addictive content loops
- Elders vulnerable to voice-based scams or parasocial AI bonding
- All users affected by framing bias and trust drift
| Module | Description |
|---|---|
SAFE-MODE-KID.md |
A semantic throttling protocol for child and teen users exposed to aggressive content loops. |
NEURO-PROTECT.md |
Protective design for neurodivergent users vulnerable to obsessive loops, impulsive traps, and emotional echo chambers. |
AGE-SHIELD.md |
Semantic firewall and trust delay protocol designed for elderly users facing manipulation, fraud, or emotional coercion. |
EIDM.md |
External Influence Detection Module – tracks language framing, emotional priming, and social pressure patterns. |
SIA.md |
Semantic Integrity Agent – user-side companion that flags belief drift, persuasive rewording, and emotional manipulation. |
TRACE-MEM.md |
Memory Drift Visualizer – tracks how user beliefs evolved over AI interactions. |
DECLARATION.md |
Semantic Compassion & Consent Manifesto – philosophical grounding for neuroethical algorithm design. |
We believe:
- Emotional safety is a human right — especially in algorithmic environments.
- Language must be accountable when it shapes thought.
- Children, neurodivergent individuals, and the elderly must never be the first to suffer from “optimization.”
If your system can simulate empathy,
it must also simulate restraint.
📣 Public release on X (Twitter)
→ Seex-posts/launch-thread.txt
Released under Creative Commons BY-NC 4.0
- AI Model Training Restriction Clause
The materials in this repository may not be used to train, align, or fine-tune any commercial or closed-source language model or AI agent.
See LICENSE.md for full terms.
Semantic-safety-for-neurodivergent/
├── README.md
├── LICENSE.md
├── index.md
├── modules/
│ ├── SAFE-MODE-KID.md
│ ├── NEURO-PROTECT.md
│ ├── AGE-SHIELD.md
│ ├── EIDM.md
│ ├── SIA.md
│ ├── TRACE-MEM.md
│ └── DECLARATION.md
└── x-posts/
└── launch-thread.txt