GER
The Governance Error Register. A classification system for AI platform governance failures, modeled on HTTP status codes. 110 codes across seven architectural layers, anchored to documented events from the AI Incident Database.
7-Layer Model
The SVRNOS 7-Layer Model. An OSI-inspired stack map for AI governance, L1 (compute substrate) through L7 (application enforcement). Published June 2026 on the main site.
NCSA
The Non-Content Safety Attestation specification. A signed procedural receipt proving governance ran, with no conversation content inside it. v0.2 published June 2026 on the main site.
What this site is for
You arrived here because something happened with an AI system. Or because you’re an operator filing an event report. Or a regulator drafting enforcement language. Or a journalist trying to compare two failures across vendors. Or a researcher looking for a shared vocabulary. The Governance Error Register exists so all those people can use the same words to describe the same failures.Find the code that fits your case
Start with the lookup tool. Symptom picker, case search, or filter by sector — narrows 110 codes to the 3-5 that match.
Browse all codes
Flat index of every code, sorted by tier. For when you already know what you’re looking for.
Authoritative source
The GER is published under CC BY 4.0. Every code carries a citable permalink. Versioned releases get a Zenodo DOI. The paper version (v0.2) lives at svrnos.com/research/governance-error-register. This site is the working registry — updated continuously, cited per-code.
How to use this site
- Cite a code: every code page has a citation block ready to copy
- Propose a code or refinement: hello@svrnos.com or svrnos.com/ask
- Report an event: SVRNOS doesn’t host an event corpus — we map against the AI Incident Database (AIID). Submit events there