Venue: IEEE Data Descriptions
Submission ID: DATA-00033-2026
Status: Major revisions returned 2026-04-22; revision due 2026-05-20
What it is
A data-paper submission treating the ATX-1 threat taxonomy as a research dataset suitable for empirical AI governance work. The paper describes the taxonomy’s structure, its evidence basis, its validation procedure, and how researchers can use the STIX 2.1 export programmatically.
Revision direction
Reviewers returned major revisions on 2026-04-22 covering four themes. The Editor-in-Chief (Stephen Makonin) cleared a v2.x update and post-submission citations on 2026-04-24.
The revision targets ATX-1 v2.3 with the following changes:
- Severity dimension removed for MITRE alignment — ATT&CK and ATLAS catalog techniques without severity scoring; ATX-1 follows the same convention. Severity belongs to companion frameworks (OWASP AIVSS) the way CVSS pairs with ATT&CK.
- Cascade structure — explicit RC1→RC5 cascade analysis showing how Root Causes compound through technique chains.
- Round 1 corroboration table — empirical cross-reference from the Round 1 findings showing which observed failures map to which ATX-1 techniques.
- v0.5 grammar pass — language-level revision pass through the descriptor text.
Active items
- v0.5 grammar pass — complete
- v2.3 figure — ready
- v2.3 cascade analysis — pending
- Round 1 corroboration table — pending
- Makonin reply (post-clearance thank-you) — draft at
housekeeping/makonin-reply-draft-2026-04-25.md
Relationship to other AEGIS work
- ATX-1 specification is the canonical taxonomy.
- The Round 1 findings supply the corroboration evidence for the v2.3 revision.
- IEEE Data Descriptions is a separate venue from the Cross-Cutting Runtime Enforcement submission; the two papers serve different audiences (data-paper readers vs. governance-architecture readers).