Chapter 25: AI Governance & Ethics¶

Operationalize responsible AI: an EU AI Act-style risk classifier, model card generation and validation, risk matrices, and the governance frameworks that gate deployment.

Metadata¶

Learning Objectives¶

• Classify systems by risk with an EU AI Act-style rule set (minimal → unacceptable)
• Generate model cards that document intended use, data, metrics, and limitations
• Validate documentation completeness before release
• Assess risk with a likelihood × severity matrix
• Apply fairness and transparency principles to real decisions
• Navigate regulation — the EU AI Act, NIST AI RMF, and data-protection basics
• Define accountability — owners, audits, and incident response
• Build an ethical culture — review boards, escalation, and long-term strategy

What's Included¶

Notebooks¶

Scripts¶

• config.py — Chapter config: required model-card fields, high-risk domains
• governance.py — Risk classification, model cards, validation, and risk matrices

Exercises¶

• Problem Set 1: Risk & Matrices — Classify systems into risk tiers and score a risk matrix
• Problem Set 2: Documentation — Generate a model card and validate its completeness
• Solutions — in exercises/solutions/ (notebooks and solutions.py for CI)

Diagrams (Mermaid)¶

• risk_tiers.mermaid, governance_flow.mermaid

Read Online¶

• 25.1 Introduction — An EU AI Act-style risk classifier and risk matrices
• 25.2 Intermediate — Generating and validating model cards for accountability
• 25.3 Advanced — Regulation, accountability, and building an ethical culture

Or try the code in the Playground.

How to Use This Chapter¶

git clone https://github.com/luigipascal/berta-chapters.git
cd berta-chapters/chapters/chapter-25-ai-governance-and-ethics
pip install -r requirements.txt
jupyter notebook notebooks/01_risk_classification.ipynb

Created by Luigi Pascal Rondanini | Generated by Berta AI