AI Compliance Glossary
Definitions of key terms in AI governance, EU AI Act compliance, and regulatory frameworks for compliance officers and practitioners.
Compliance
Compliance processes and documentation requirements
Evidence Pack
A comprehensive bundle of documentation, logs, and artifacts that demonstrate AI system compliance for auditors.
Model Card
A standardized document describing an AI model's intended use, performance, limitations, and ethical considerations.
Post-Market Monitoring
Ongoing surveillance of AI system performance and compliance after deployment to identify and address issues.
System Card
Documentation describing a complete AI system including its architecture, components, data flows, and operational context.
EU AI Act
Key terms and concepts from the EU Artificial Intelligence Act
AI literacy
AI literacy is the skills, knowledge and understanding people need to deploy AI systems in an informed way and to be aware of their opportunities, risks and possible harms.
Annex III
The EU AI Act annex listing categories of high-risk AI systems subject to strict compliance requirements.
Annex IV
The EU AI Act annex specifying technical documentation requirements for high-risk AI systems.
CE Marking
A certification mark indicating that an AI system complies with EU health, safety, and environmental protection requirements.
Conformity Assessment
The process of evaluating whether an AI system meets all applicable EU AI Act requirements before market placement.
Deployer
An organization that uses an AI system under its authority, except for personal non-professional use.
FRIA
Fundamental Rights Impact Assessment - an evaluation of how an AI system may affect fundamental rights.
High-Risk AI System
An AI system subject to strict requirements under the EU AI Act due to its potential impact on health, safety, or fundamental rights.
Notified Body
An organization designated by an EU member state to assess conformity of high-risk AI systems.
Provider
An entity that develops or has an AI system developed and places it on the market or puts it into service.
AI Governance
Core AI governance concepts and frameworks
Agent authorization
Agent authorization is the check, made each time an AI agent proposes an action, that decides whether the agent may take that action with those arguments for its principal.
Agent Mandate
An agent mandate is an explicit, machine-readable definition of the authority a principal delegates to an AI agent.
AI agent registry
An AI agent registry is an organization's inventory of its AI agents, recording for each one its owner, purpose, version, permissions and status.
AI Governance
The operational layer that defines who can approve AI decisions, what documentation must exist, how oversight is maintained, and what evidence is captured for accountability.
Disposition Engine
A disposition engine evaluates every proposed agent action against an institution's controls and returns one binding outcome before the action executes.
Governance Envelope
A governance envelope packages a proposed agent action with the full context needed to assess it before execution.
Human Oversight
Human oversight is the ability of assigned, competent people to monitor an AI system while it is in use, interpret its output, and override, reverse or stop it.
Policy-Bound Execution
Policy-bound execution evaluates every proposed agent action against a published, versioned policy before the action runs.
Runtime Governance Layer
A runtime governance layer sits between an AI agent and the systems it acts on, evaluating each proposed action before execution.
Tamper-Evident Lineage
Tamper-evident lineage is an append-only record of an agent's actions and the governance decisions applied to them, whose integrity can be verified independently.
Technical
Technical concepts for AI system monitoring and control
Agent kill switch
An agent kill switch is a control that stops an AI agent's active runs, refuses its next actions and revokes its access in one command.
Audit Trail
A chronological record of AI system activities, decisions, and human interactions that enables traceability and accountability.
Bias Detection
The process of identifying and measuring unfair or discriminatory patterns in AI system outputs or training data.
Data provenance
Data provenance is the record of where a piece of data came from, what produced it, and how it changed before it was used.
Deterministic replay
Deterministic replay is re-running a recorded AI agent decision with the same inputs, policy version and configuration to confirm that it produces the same outcome.
Drift Detection
Monitoring AI system performance over time to identify degradation or deviation from expected behavior.
Explainability
Explainability is the degree to which people can understand how an AI system reached an output or decision and explain it to others.
Guardrails
Technical constraints and policy controls that prevent AI systems from producing harmful or non-compliant outputs.
Ready to implement AI governance?
See how KLA helps regulated organizations operationalize these concepts with policy checkpoints, audit trails, and evidence exports.