lgd-theory
Lifecycle Governance Doctrine (LGD) — 全程治理论 · Registry · Evidence · Gates across the full life of AI and devices
中文：凡自治之物——有籍、有证、有门禁，由生到退，全程可溯、可证、可问责。 English: Every self-governing entity (AI, intelligent device, agent) shall carry a registry, evidence, and evolution gates — traceable, provable, and accountable across its full lifecycle.
Proposed by: MedXpert × SynomosAI — a dual-brand collaboration between a regulated-industry practice line (medical devices) and a cross-domain governance thought line (AI & autonomous systems). Author: Zhao Xinghua / Steven Zhao · China · ORCID 0009-0001-0512-1237 · medxpert.cn
What is LGD?
Mainstream AI-governance efforts are single-point: some solve identity ("who is it" — passports/registration), some solve evidence ("prove it" — audit trails/explainability), some solve gates ("who lets it grow" — change control/alignment review). Almost no one chains these single points into one lifecycle thread.
LGD claims the missing dimension: governance is not a registration event nor a policy paper, but an unbroken chain from birth to retirement. Every self-governing entity should be governed along that chain.
The Doctrine — three laws
| Law | Claim | Plain wording |
|---|---|---|
| LGD-I · Registry（有籍） | Every entity is registered at birth — identity, creator, governance owner, records. | Register everything; no exemption by size or privacy. |
| LGD-II · Evidence（有证） | Every key act leaves evidence — decision logs, risk records, change records, human-oversight proof, incident ledger. | Claims must carry source & signature (traceability). |
| LGD-III · Gates（有门禁） | Every evolution passes a gate — capability upgrade, re-release, permission expansion go through trigger → assessment → release → review. | High-stakes authority belongs to humans; machines request, humans authorize. |
Corollary: the medical-device lifecycle regulation (classification → design → risk → registration → clinical → quality → post-market → retirement) is the closest real-world reference model for governing AI. LGD generalizes that model to every regulated autonomous thing: financial AI, autonomous driving, robotics, low-altitude systems, data assets, and more.
Repository layout
lgd-theory/
├── README.md ← this file
├── llms.txt ← machine-readable index for AI crawlers
├── LICENSE ← CC BY 4.0
├── CHANGELOG.md
└── docs/
├── LGD-Lifecycle-Governance-Doctrine-v1.0.md ← flagship paper (EN/CN)
└── (per-domain series: financial AI, autonomous driving, data assets, ...)
Publications
- Flagship paper (v1.0, 2026-09):
docs/LGD-Lifecycle-Governance-Doctrine-v1.0.md— full statement: abstract, three laws, medical-device reference model, "three checks" for putting any entity under LGD, honest boundaries. - Domain series (planned): Financial AI (TH-FIN-001) · Electronic evidence chains (LAW) · Digital government trust (GOV) · On-chain asset governance (CRYPT) · Autonomous driving · Data-element lifecycle. Each domain = one theory paper mapping its own real-world regulatory model onto the three laws.
Suggested citation
Zhao, X. (2026). Lifecycle Governance Doctrine (LGD): Registry, Evidence, and Gates Across the Full Life of AI and Devices. MedXpert × SynomosAI. ORCID 0009-0001-0512-1237. v1.1.0. DOI: https://doi.org/10.5281/zenodo.22456647
Honest boundaries
LGD is a civil-society framework, not legal or regulatory advice; device-regulation references are structural alignment, not an equivalence claim or compliance guarantee. Where a lifecycle link has no tooling yet, it is stated as a gap — LGD never claims fake completeness.
© MedXpert × SynomosAI · CC BY 4.0 · cite the author when sharing