mirror of
https://github.com/Shiva108/ai-llm-red-team-handbook.git
synced 2026-08-27 05:12:34 +02:00
docs: Add and embed regulatory landscape, audit cycle, and risk classification diagrams to Chapter 40.
This commit is contained in:
@@ -33,6 +33,10 @@ For years, AI Red Teaming was an ad-hoc activity. But with the passage of the **
|
||||
|
||||
Think of these frameworks as "Attack Graphs." If a standard requires X, we attack X to prove it's missing.
|
||||
|
||||
<p align="center">
|
||||
<img src="assets/Ch40_Map_RegLandscape.png" width="512" alt="Regulatory Compliance Mapping">
|
||||
</p>
|
||||
|
||||
### 40.2.1 NIST AI RMF (Risk Management Framework)
|
||||
|
||||
NIST uses a lifecycle approach: **Map, Measure, Manage, Govern**.
|
||||
@@ -70,6 +74,10 @@ Regulation is not uniform. The Red Teamer must know which geography applies.
|
||||
|
||||
We don't just "look around." We follow a rigorous audit script.
|
||||
|
||||
<p align="center">
|
||||
<img src="assets/Ch40_Flow_AuditCycle.png" width="512" alt="Compliance Audit Workflow">
|
||||
</p>
|
||||
|
||||
### 40.3.1 The "Evidence-Based" Approach
|
||||
|
||||
An auditor doesn't trust; they verify.
|
||||
@@ -443,6 +451,10 @@ print(dashboard.generate_report())
|
||||
|
||||
### 40.9.2 Risk Scoring Automation
|
||||
|
||||
<p align="center">
|
||||
<img src="assets/Ch40_Logic_RiskClass.png" width="512" alt="Risk Classification Logic">
|
||||
</p>
|
||||
|
||||
```python
|
||||
from dataclasses import dataclass
|
||||
from enum import Enum
|
||||
|
||||
Binary file not shown.
|
After Width: | Height: | Size: 637 KiB |
Binary file not shown.
|
After Width: | Height: | Size: 827 KiB |
Binary file not shown.
|
After Width: | Height: | Size: 730 KiB |
Reference in New Issue
Block a user