AI Risk Map v1.3

12 domains · 55 categories · 196 risk factors. A content release: eight new factors, one new category, a wider scope statement, and re-tested cross-references.

If you used v1.2, nothing you cited has disappeared: identifiers, columns, and file names are unchanged. The full list is on the workbook's Change log sheet.

New to leverage

• Eight factors. Agent actions third parties can't attribute to you (AIRM-190); safety training stripped or eroded in a self-hosted model (AIRM-191); changes to your own AI service made without notice to customers (AIRM-192); your advantage eroding as competitors and customers approximate it with AI (AIRM-193); AI agents disintermediating you from your customers (AIRM-194); leaders managing AI like a conventional IT rollout (AIRM-195); insiders using AI agents to reach beyond their own access (AIRM-196); inputs crafted to evade AI detection models (AIRM-197).
• A new category, Competitive position, in Strategy & Value Realization: AIRM-193, AIRM-194, and AIRM-026 (moved, wording unchanged).
• A wider scope statement. Only others' adversarial or malicious use of AI is excluded. How competitors' and customers' AI changes the value of your own strategy is now in scope.
• More see-also pointers (54, up from 20), so related risks are easier to read together.

Check against earlier work

• Four factors narrowed. If you mapped controls or scenarios to these, some may belong elsewhere now:
  • AIRM-149 covers abandoned open-source dependencies only. Known vulnerabilities left unpatched are AIRM-079.
  • AIRM-114 covers deliberate denial-of-wallet only. Unmanaged usage of your own systems is AIRM-164.
  • AIRM-122 covers drift from uncontrolled learning only. Deliberate poisoning through a learning loop is AIRM-065.
  • AIRM-153 covers workforce literacy only. Leadership understanding is the new AIRM-195.
• One factor clarified to its cause. AIRM-161 now means customers distrusting or avoiding AI-mediated services. If you used it for loss of trust after an incident, use AIRM-160.
• One merge. AIRM-171 is merged into AIRM-165. It stays in the CSV and workbook as a retired row pointing to AIRM-165.
• Fewer split pairs, so counts may change. Every v1.2 pair was re-tested against the counting question above. Six remain (14 marked factors, down from 32). The rest are now see-also pointers, which count both. If you counted a v1.2 pair once, check whether it is still a pair.
• Sources. DASF 5.4 (malicious libraries) moved from AIRM-079 and AIRM-149 to AIRM-148. The ATLAS evasion citations moved from AIRM-057 and AIRM-058 to the new AIRM-197. MAS Guidelines on AI Risk Management are a new source.
• Wording. Fourteen other factors were reworded to widen or clarify the same condition. Cite by ID and these need no action.
• Terminology. "Two-faced pair" is now "split pair."

Files: poster (PDF, PNG, SVG), workbook, CSV, and markdown corpus, all under the same file names as v1.2. Cite this version as blob/v1.3/…; link to blob/main/… for the current version.

AI Risk Map v1.2

Clarifications and cross-reference corrections from external review. The risk content is unchanged: same 12 domains, 54 categories, and 189 risk factors, with every identifier the same as v1.0.

What's new

Control framing. The map is not a control standard or control library. Many factors imply a missing or deficient control condition, because a control gap is a condition that contributes to risk. But no factor is phrased as a control to implement, and the map does not say which control closes a gap: one factor may call for several controls, and one control may address several factors.

Cross-reference markers. »n now means one thing: the other face of this risk is owned by domain n, so count the pair once.

• Both faces of all 15 two-faced pairs are now marked. v1.1 marked both faces of only four pairs and left three unmarked.
• Related-but-different risks that shared the same marker are now listed as see-also pointers on a new part of the workbook's Cross-references sheet (20 pointers, by factor ID). They no longer appear on the poster.
• Net result: 32 markers on 30 factors, down from 35.

Corrections. The Transparency pair listed AIRM-092 in error; it is now AIRM-093.

For AI-assisted use. The markdown corpus now carries the map's design principles (scope, retention gates, factor form, two-faced rule, markers), and each marked factor names its partner IDs.

Each change is recorded by factor ID on the workbook's Change log sheet.

Linking to the map

• Always the latest version: https://github.com/wheels344i-gte/ai-risk-map/blob/main/AI-Risk-Map.pdf
• This version exactly: https://github.com/wheels344i-gte/ai-risk-map/blob/v1.2/AI-Risk-Map.pdf

Swap AI-Risk-Map.pdf for any file below. For tools, scripts, or embedding an image, replace blob with raw. The v1.0 files remain attached to the v1.0 release.

Files

All files are in the repository at this version.

File	Use
AI-Risk-Map.pdf	Print-ready poster, 48×36 in
AI-Risk-Map.png / .svg	Poster, raster and vector
AI-Risk-Map-preview.png	Readable detail for posts and slides
AI-Risk-Map-Taxonomy.xlsx	Every factor with sources, cross-references, scope decisions, and change log
AI-Risk-Map.md	Full corpus for loading into your own AI assistant
AI-Risk-Map.csv	Flat table for mapping controls or tools, frozen nine-column schema

Licensed CC BY-SA 4.0.

AI Risk Map v1.1

A design refresh with a small set of cross-reference corrections. The risk content is unchanged: same 12 domains, 54 categories, and 189 risk factors, with every identifier the same as v1.0.

What's new

Poster redesign. New typeface (Figtree), a lighter header with a how-to-read key, categories set as colored text, and cross-reference markers drawn as pills in the target domain's color.

Cross-reference corrections.

• Removed two markers that pointed at their own domain (AIRM-046, AIRM-152).
• AIRM-187 now shows both of its targets, Governance (»1) and Legal (»8). The v1.0 poster printed only the first.
• Added the Governance / Autonomy accountability pair (AIRM-006 / AIRM-187) to the Cross-references sheet.

Each change is recorded by factor ID on the workbook's Change log sheet.

Stable file names. File names no longer carry a version number, so links keep working across releases. The version is printed on the poster, on the workbook's Version History sheet, and in the markdown header.

Linking to the map

• Always the latest version: https://github.com/wheels344i-gte/ai-risk-map/blob/main/AI-Risk-Map.pdf
• This version exactly: https://github.com/wheels344i-gte/ai-risk-map/blob/v1.1/AI-Risk-Map.pdf

Swap AI-Risk-Map.pdf for any file below. For tools, scripts, or embedding an image, replace blob with raw. The v1.0 files remain attached to the v1.0 release.

Files

All files are in the repository at this version.

File	Use
AI-Risk-Map.pdf	Print-ready poster, 48×36 in
AI-Risk-Map.png / .svg	Poster, raster and vector
AI-Risk-Map-preview.png	Readable detail for posts and slides
AI-Risk-Map-Taxonomy.xlsx	Every factor with sources, cross-references, scope decisions, and change log
AI-Risk-Map.md	Full corpus for loading into your own AI assistant
AI-Risk-Map.csv	Flat table for mapping controls or tools, frozen nine-column schema

Licensed CC BY-SA 4.0.

AI Risk Map v1.0

First public release. 12 domains, 54 categories, 189 risk factors, each with a permanent identifier (AIRM-001 onward). Structure: organized by the enterprise function that owns the risk, not by AI lifecycle. Scope: risk arising from the organization's own adoption of AI. Validation before release: crosswalked against fifteen frameworks and regulations; persona and named-expert critique; five rounds of external practitioner review; used as the validation standard for an independent 264-control catalog; and a full citation audit in which every named source locator was verified against the primary text and every factor tagged Original was checked against every cited framework. Result: 189 of 189 factors traceable; 98 carry the author's framing; 30 are named by no reviewed framework. From this version, changes are recorded at factor grain on the Change log sheet, and identifiers never change.