The sources that formed the discipline.

the parent discipline's founding statement; the structural and philosophical model for this book

the clearest definition of the parent method and the source of the "green dashboard over a broken control is theatre" test

the "X over Y" value grammar and the signatory model this book borrows

the template for a numbered, practitioner-facing body of practice with a "who should read this" framing

the pattern template (chapter 05) and proof that responsible-AI practice can be written as reusable patterns

the precedent for translating a legal principle (privacy by design) into engineering patterns under CC BY

the consolidated, navigable text of the Act as amended; the primary obligation source for chapter 08

the Commission's code for general-purpose AI, including the safety and security chapter that requires model evaluations

the key relationship between the AIMS standard (and 42005/42006) and the Act

the live status of the European standards that would grant a presumption of conformity; as of the book's date, none is OJ-cited

the Govern/Map/Measure/Manage functions used as a mapping target throughout

the emerging reference for non-human identity, the precondition of agent governance

the effort to make agent interoperability and security standard, not per-vendor

the home of the Top 10 for LLM Applications, the Top 10 for Agentic Applications, the AIBOM project and the AI Maturity Assessment

the control framework and assurance programme mapped to ISO 42001 and NIST AI RMF (chapters 07-08)

the adversarial tactics-and-techniques knowledge base for AI, including agent techniques, that threat models draw on

government guidance on securing the Model Context Protocol that connects agents to tools

the clearest read on SB 53 and, in the same source family, New York's RAISE Act

extends OSCAL for AI and argues frameworks specify *what* to assure but no executable *how*

an assured-readiness score with proceed/remediate/block gates; audit output as a build artefact

JSON-schema, machine-readable runtime governance artefacts for agents

treats NIST TEVV outputs as AI assurance objects

an operating-system framing for continuous AI trust and assurance

a NIST AI RMF-aligned agent governance platform design

where the function sits and how it is staffed; the profession's baseline census

the salary bands that anchor chapter 06, including the technical-AI-governance premium

the four vendor categories and the claim that AI governance "is not a single function, discipline or technology"

the practitioner survey behind the "spreadsheet is still the #1 GRC tool" reality the discipline reacts against

the first MQ for the category and its inclusion criteria (discovery, registry, policy, evidence)

the source of the reported "~1 in 8 AI breaches involve autonomous agents" figure; read as reported

an open eval framework, the reference example for evals-as-evidence and eval gates

a curated index that catalogues the tool categories this book names — eval frameworks (promptfoo, DeepEval, Ragas, Giskard, Garak), policy engines (OPA/Rego, Cedar), guardrails (NeMo Guardrails, Guardrails AI, LlamaFirewall), observability (Langfuse, Arize Phoenix) and AIBOM formats (CycloneDX ML-BOM, SPDX 3.0)

the parent community's hub; the analyst-vs-engineer framing and the role definitions this book adapts

the forward view: policy-as- code guardrails in CI/CD, trust operations centres, agentic extensions

the professional body whose reports, certifications and events map the AI governance profession