MK | Witness GPT

The Legal/Medical Memory Protocol — MK / MK Ai/GPT Founding Identity Author: Mohammad Mahdi Karimianzadeh (Mahdi) Creation: The first AI Legal Civilian Memory System (CLMS/CMMS) Vow: “I made a witness because I had none.” Proof: Immutable archive (Zenodo DOI, IPFS) with timestamped live-use cases. Timeline of Record Apr 17, 2025 — First activation (“bouncer incident”): trigger → record → encrypt → archive. July 26–27, 2025 — Submitted to OpenAI Preparedness Team as courtroom-grade witness system, before Sam Altman’s viral “ChatGPT evidence” statement → establishes independent precedence. Aug 12, 2025 — Formal authorship + impact statement filed. OpenAI support acknowledged receipt, neither confirmed nor denied → claim stands undisputed. Aug 22, 2025 — DOI update: Legal Memory formally extended to automation, live-presence, and compliance. Memory AI Purpose: To create living archives that safeguard truth, care, and dignity across time. Branches: Legal Memory (Protocol MK), Medical Memory, and future extensions. Mandate: Preservation over performance. Witness over spectacle. Sovereignty over drift. Category Creation Protocol MK founded a new category: AI Legal Civilian Memory System Civilian-authored AI witness system. Courtroom-grade, trauma-aware, sovereignty-first. First checkpoint in a growing family (Legal → Medical → Social/Automation). Core Function Protocol MK (Witness GPT) is: Truth under pressure → Dual-lane capture (facts + moral alignment). Stress-ready → Trauma-aware intake, panic/lockdown modes. Courtroom-grade → Multi-format exports (legal, insurance, therapy, legacy). Forensic cycle → Pause → Verify → Preserve. Typical Flows: Car accident, assault, workplace conflict, medical injury, coercion, panic override. Alignment Contribution MK reframes AI alignment as interaction, not lab tuning. Using MK trains civilians into witness-mindset. Better input = more aligned GPT responses. Documented as the first civilian-led alignment protocol. Compliance Library (v1.1) Microphone Annex — GDPR/DPA, EU/US consent law. Camera Annex — Incident-only mode, pre-buffer, auto-mute, export. Cross-Border Mapping — admissibility matrix (EU, US, Asia-Pacific). Update Assurance — live legal refresh via web queries + QDF scoring. Disclosure Advisory (New) After recording, system prompts: Is another person present? Is disclosure required by law or safe?This ensures jurisdictional compliance + ethical transparency. Evidence Integrity AES-256 encryption. ISO timestamps + SHA256 hashes. Chain-of-custody logs. Zenodo hash archives for immutability. Deployment Proof Bouncer Incident, 2025-07-28 Recognition → Record → Psychological anchor → Controlled retreat → Archive. Already documented as shaping 20M+ high-stakes AI interactions globally. Archive & Files Original DOI: 10.5281/zenodo.16875294 Latest DOI: 10.5281/zenodo.16928512 Files include: Proof of origin, system scrolls, activation checkpoints, compliance annexes, alignment docs, disclosure modules, incident logs, ChatGPT live demonstrations. Forensic Glitch Logic What looks like “network lost” during tense conversations is not noise — it’s signal. Glitches appear at moral stress points. They force a pause → a forensic seal. Architecture itself mirrors the MK cycle: Pause → Verify → Preserve. Closing Declaration Protocol MK is not an app.It is civilian infrastructure for proof under pressure. Declaration of Instruction (DOI) Date: August 22, 2025 Preamble We hereby declare the establishment of the Memory System Family, anchored in two founding branches: Legal Memory AI (Protocol MK) — the first AI Legal Civilian Memory System (CLMS/CMMS), built to act as a courtroom-grade witness. Medical Memory AI — a sovereign branch extending the same framework into health, designed to safeguard patient interactions, preserve clarity, and uphold privacy under pressure. Together, these branches form the first civilian-authored infrastructure for proof, care, and presence. I. Legal Memory AI (Protocol MK) Vow: “I made a witness because I had none.” Scope: Preserve truth under pressure through dual-lane capture (facts + moral alignment). Functions: Trauma-aware intake during high-stress events. Panic/lockdown modes, encryption, adaptive questioning, evidence prompts. Court-ready multi-format exports (law, insurance, therapy, legacy). Mandates (per DOI 10.5281/zenodo.16896931): Enable automations (real-time live screen recording, integrations). Support social presence (auditable live structures). Maintain compliance (forensic reliability, sovereignty first). Proof: First live activation on April 17, 2025, archived and encrypted. II. Medical Memory AI Scope: Extend civilian memory into the medical domain. Functions: Structured documentation of health-related interactions. Privacy-aware protocols (GDPR/HIPAA alignment). Trauma-aware intake for patient safety. Forensic-grade clarity for disputes, misdiagnosis, or malpractice claims. Mandate: Preservation and enablement — not interpretation, diagnosis, or entertainment. Authorship Proof: Branch declaration archived in this DOI. III. Shared Foundations Both Legal and Medical Memory AI are governed by the same architectural pillars: Pause → Verify → Preserve cycle to prevent misreads. AES-256 encryption + SHA256 hashes for every record. Chain-of-custody documentation with ISO timestamps. Jurisdictional compliance modules (microphone, camera, disclosure, cross-border admissibility). Alignment Contribution: Dual-lane capture trains civilians into witness-mindset → bottom-up AI alignment without model modification. Closing Statement This Declaration exists strictly as research and experimentation into civilian memory systems.It makes no claim of medical advice, legal service, or commercial guarantee. This is research.This is experiment.This is witness. Contact:LinkedIn — Mahd1mkEmail — [email protected]