Glossary

AI governance glossary

The AI governance glossary: the words procurement uses, defined short, each with the primary source behind it. No vendor glossaries, no marketing terms.

Terms

Agentic workflow: These are AI agents that plan, call tools and decide what happens next.; Source: arxiv.org
AI governance: These are the decisions and controls that manage AI risk.; Source: nist.gov
AI impact assessment: AI impact assessment; Source: eur-lex.europa.eu
AI management system: AI management system; Source: iso.org
AI operating model: AI operating model; Source: hkdca.com
AI register: This is a company's inventory of every AI system it uses.; Source: nist.gov
Conformity assessment: Conformity assessment; Source: eur-lex.europa.eu
Data classification: Data classification; Source: csrc.nist.gov
DPIA: DPIA; Source: eur-lex.europa.eu
EU AI Act: This EU regulation sorts AI systems into risk tiers.; Source: eur-lex.europa.eu
Evals: These are repeatable tests that score an AI system's output.; Source: nist.gov
Guardrails: Guardrails; Source: genai.owasp.org
Hallucination rate: Hallucination rate; Source: nist.gov
High-risk AI system: High-risk AI system; Source: eur-lex.europa.eu
Human oversight: Human oversight; Source: eur-lex.europa.eu
Human-in-the-loop: A person reviews or stops the output before it lands.; Source: eur-lex.europa.eu
ISO/IEC 42001: This certifiable standard sets how a company governs its AI.; Source: iso.org
LLM gateway: Every model call passes through one gateway, for control.; Source: arxiv.org
Model card: Model card; Source: arxiv.org
Model Context Protocol: Model Context Protocol; Source: modelcontextprotocol.io
Model provenance: Model provenance; Source: eur-lex.europa.eu
NIST AI RMF: This voluntary US framework helps manage AI risk.; Source: nist.gov
Observability: Observability; Source: nist.gov
Post-market monitoring: Post-market monitoring; Source: eur-lex.europa.eu
Prompt injection: Prompt injection; Source: genai.owasp.org
Provider vs deployer: Provider vs deployer; Source: eur-lex.europa.eu
Red teaming: Red teaming; Source: nist.gov
Retrieval augmented generation: Retrieval augmented generation; Source: genai.owasp.org
Serious incident reporting: Serious incident reporting; Source: eur-lex.europa.eu
Shadow AI: These are AI tools staff use for work without security's approval.; Source: nist.gov
Sub-processor: Sub-processor; Source: eur-lex.europa.eu
Technical documentation: Technical documentation; Source: eur-lex.europa.eu