TA-14 GOVERNED KNOWLEDGE SYSTEM
AI Governance Dictionary
Navigate the roles, evidence concepts, assurance methods, risk classifications, documentation duties, and execution principles used across AI governance. Each entry separates general terminology from the TA-14 admissible-execution interpretation.
TERMINOLOGY CONTROL DESK
Find the term. Inspect its boundary.
Definitions can change by jurisdiction, authority, version, system role, and use context. Dictionary entries support navigation; the controlling source remains authoritative.
KNOWLEDGE DISTRIBUTION
Browse by governance function.
DICTIONARY RESULTS
22 terms available
The assignment and preservation of responsibility for AI decisions, controls, evidence, and outcomes.
- EU AI Act
- NIST AI RMF
- ISO/IEC 42001
- OECD AI Principles
Authority record
Approval record
Audit record
Outcome record
Accountability must remain bound to a valid authority and a preserved execution record.
The condition under which evidence, authority, continuity, and execution requirements are sufficient for a consequential action to proceed.
A structured basis for confidence that an AI system, process, or control satisfies defined requirements.
The valid and attributable basis by which an actor, system, reviewer, or institution is permitted to make or approve a consequential decision.
A systematic tendency that can produce unfair, distorted, or unrepresentative AI behavior or outcomes.
The act of connecting a proposed decision or execution to the exact identity, evidence, authority, conditions, and controls used to evaluate it.
A process used to determine whether a system, product, or organization satisfies specified requirements.
The preserved connection between the system, evidence, authority, controls, decision state, execution, and outcome over time.
An entity that uses an AI system under its authority, excluding purely personal, non-professional use.
A material change in data, system behavior, operating conditions, controls, or outcomes that may invalidate prior evidence or determinations.
Recorded information used to support a claim, determination, authority, control, decision, execution, or outcome.
The extent to which the reasons, factors, or logic behind an AI output can be meaningfully communicated.
A broadly capable model trained on extensive data and adaptable to many downstream tasks.
An AI model capable of performing a broad range of tasks and being integrated into multiple downstream systems.
An AI use or system category subject to heightened governance obligations because of its potential effects on safety or fundamental rights.
Human involvement designed to understand, supervise, interrupt, override, or otherwise govern AI operation.
Preserved evidence showing what happened after a governed action was executed and whether the intended control or intervention worked.
The ongoing collection and evaluation of information about system performance after deployment.
The attributable history of where evidence, data, documentation, or a decision originated and how it changed over time.
An entity that develops an AI system or general-purpose AI model and places it on the market or puts it into service under its own name or trademark.
Structured records describing a system's design, intended purpose, capabilities, limitations, controls, testing, and changes.
Evidence that a system is suitable for its intended use and performs acceptably in the relevant context.
TERMINOLOGY GOVERNANCE SEQUENCE
Definitions inform the route. They do not control it alone.
Locate the governing term and its apparent source context.
Separate similar terms, roles, evidence types, and legal effects.
Review the controlling authority, version, jurisdiction, and scope.
Determine whether the term governs the declared entity and use.
Bind the meaning to the evidence, authority, decision, and action.
Preserve the interpretation used at decision and execution time.
TERMINOLOGY BOUNDARY
A definition is not a legal or execution determination.
Dictionary entries organize governance language and explain TA-14 admissible-execution implications. They do not independently establish applicability, conformity, compliance, certification, authority, admissibility, or permission to execute. The controlling source, jurisdiction, version, role, evidence, and declared operating context must still be reviewed.