Skip to content

What AI-LinkMO adds to AI Atlas Nexus

Explanationrationaleprovenanceexplanations/what-ai-linkmo-adds

Checked by automation only

still a draft

Written 2026-10-17 by ktl-librarian (automation). Last checked 2026-10-27 by ktl-librarian (automation).

The full record
Checks
  • 2026-10-27 by ktl-librarian (automation)
Sources
References
Related to
Graph
Open this concept in the knowledge graph
IRI
https://noelmcloughlin.github.io/ai-linkmo/knowledge/explanations/what-ai-linkmo-adds
Licence
CC BY 4.0
Files
MarkdownJSON-LDSource on GitHubHistory on GitHub

Why linked AI governance says why the model is shared. This concept says whose model it is, and what this repository adds to it.

IBM AI Atlas Nexus provides three things. The AI Risk Ontology is the LinkML schema every CLI command, API route, UI form and graph label here is generated from. The packaged risk data carries the IBM AI Risk Atlas, its public taxonomies and the relationships between them. The related-risk, related-control and related-incident lookups, and so the crosswalks, follow those relationships through the ai-atlas-nexus library. The library loads the ontology and the data, and LinkML’s generators produce the JSON Schema the web UI includes.

This repository adds doors, encodings and checks

Section titled “This repository adds doors, encodings and checks”

AI-LinkMO is a workshop built on those, with no ontology of its own. It adds:

  • the four access patterns (CLI, API, web UI, graph);
  • nine governance frameworks encoded in the ontology’s own classes as bring-your-own-data examples under byo/data/;
  • crosswalks on screen and as files;
  • a scripted walk through all four doors in byo/notes/EXAMPLES.md;
  • most of the CLI examples, run as tests in API and local modes;
  • curate mode, where a person adds, edits or deletes records in their own files through the web UI and PUT /byo.

The governance records say what a risk or a control is, not who vouched for the encoding. Each byo/data/ file names in a header comment the script and input that produced it, and every row carries the same dateCreated, the day the generator ran. Nothing records who checked an encoding against the framework’s own text, from which edition, or when to look again, and the ontology has no field for it. Curate mode has the same gap: the persona a person picks is not an identity, and an edited record keeps no trace of who made the change. For an auditor that is the first question, and for all nine frameworks the answer today is that nobody has checked.