Glossary
Access Management Glossary
Plain-language definitions for human and AI agent access management terms, from access reviews and SCIM to MCP gateways and zero standing permissions.
Browse Glossary Terms
All Terms
56 of 56 terms
- Protocol
A2A (Agent-to-Agent Protocol)
The Agent-to-Agent protocol (A2A) is an open standard developed by Google and the Linux Foundation that defines how AI agents delegate work to other AI agents. It provides a structured way for agents to discover capabilities, negotiate tasks, and exchange results, including a built-in state for pausing to request human authorization.
Read definition - Security
ABAC (Attribute-Based Access Control)
Attribute-based access control (ABAC) is an authorization model that evaluates access requests against attributes of the user, the resource, the action, and the environment. Instead of fixed roles, policies reference properties like department, clearance level, time of day, or data sensitivity to make dynamic access decisions.
Read definition - Concept
AI Agent Access Control
AI agent access control is the enforcement of authorization policies on actions taken by AI agents. It determines, in real time, whether a specific agent is allowed to perform a specific action on a specific resource, based on the identity of the delegating user, the properties of the action, and the policies of the organization.
Read definition - Concept
AI Agent Governance
AI agent governance is the discipline of managing how AI agents identify themselves, what they can access, and how they behave when acting autonomously on behalf of humans. It covers the policies, enforcement mechanisms, and audit trails that ensure agents operate within boundaries set by the organization, not just the user who triggered them.
Read definition - Security
API Key
An API key is a unique identifier used to authenticate requests to an application programming interface. API keys are commonly used for programmatic access, rate limiting, and usage tracking. They are the simplest form of machine authentication but carry significant security risks when mismanaged.
Read definition - Security
Access Control
Access control is the security discipline of regulating who or what can view, use, or modify resources in a computing environment. It encompasses the policies, mechanisms, and technologies that determine what actions an authenticated identity is authorized to perform.
Read definition - Process
Access Review
An access review is a systematic process of validating that existing access permissions are still appropriate for each user. Reviewers (typically managers or app owners) evaluate whether each person still needs the access they have and revoke permissions that are no longer justified.
Read definition - Concept
Agent Access Management
Agent access management is the practice of controlling what apps, data, and actions AI agents can access when acting on behalf of a user. It is the enforcement layer that sits between an agent's intent and the systems it interacts with, evaluating every request against policies defined by the organization.
Read definition - Concept
Agentic Security
Agentic security is the emerging discipline of securing AI systems that act autonomously on behalf of users. It encompasses the tools, practices, and architectures required to protect organizations from the risks introduced when AI agents access company systems, make decisions, and take actions without continuous human oversight.
Read definition - Architecture
Architecture A vs Architecture B
Architecture A and Architecture B are two enforcement approaches for AI agent governance, defined by the AARM framework (Herman Errico, 2026). Architecture A places a protocol gateway between agents and their tools. Architecture B instruments governance inside the agent runtime. Each has structural tradeoffs in coverage, deployability, and visibility.
Read definition - Process
Audit Trail
An audit trail is a chronological record of system activities that provides documentary evidence of the sequence of actions taken by users, agents, or systems. In agent governance, an audit trail captures every tool call, policy evaluation, authorization decision, and data access event for compliance and forensic purposes.
Read definition - Security
Blast Radius
Blast radius is the total extent of damage that can result from a security incident, measured by the systems, data, and operations that could be affected. In the context of AI agents, blast radius describes how far an agent could reach if its session were compromised or its behavior became unintended.
Read definition - Concept
Compositional Risk
Compositional risk is the security exposure that emerges from an agent's accumulated access path, evaluated in aggregate rather than as individual requests in isolation. Granting email access is routine when the agent only holds read-only document access. The same email access becomes an exfiltration channel when the agent already holds PII database access.
Read definition - Concept
Consent Fatigue
Consent fatigue is the desensitization that occurs when users are asked to approve too many individual actions, leading them to rubber-stamp approvals without reading them. In the context of AI agents, it is the predictable failure mode of per-action confirmation prompts used as a substitute for policy-based governance.
Read definition - Architecture
Credential Mediation
Credential mediation is an architectural pattern in which AI agents never see or hold real authentication tokens. Instead, a gateway or proxy injects credentials into outbound requests at the moment of execution. The agent holds only an opaque, session-scoped identifier that has no value outside the gateway.
Read definition - Process
Data Classification
Data classification is the process of categorizing data based on its sensitivity level and the impact of unauthorized disclosure. Common tiers include public, internal, confidential, and restricted. Classification determines which security controls, access policies, and handling procedures apply to each data set.
Read definition - Threat
Data Exfiltration
Data exfiltration is the unauthorized transfer of data from an organization's systems to an external location. In AI agent contexts, exfiltration risk arises when an agent with access to sensitive data also has the ability to communicate externally, creating a channel through which data can be extracted.
Read definition - Compliance
Data Processing Agreement (DPA)
A data processing agreement is a legally binding contract between a data controller and a data processor that governs how personal data is handled. Under GDPR, a DPA is required whenever an organization shares personal data with a third-party service provider.
Read definition - Concept
Dynamic Agent Context
Dynamic agent context is an access model in which AI agents receive the specific tools, data access, and permissions they need at the moment they are needed, scoped to the active task, and released on completion. It describes the full lifecycle of how an agent's access changes in real time, from task start to task end.
Read definition - Compliance
EU AI Act
The EU AI Act is the European Union's comprehensive regulation on artificial intelligence, published in the Official Journal in July 2024. It classifies AI systems by risk tier (unacceptable, high, limited, minimal) and imposes graduated compliance obligations on providers and deployers, with high-risk system requirements becoming fully enforceable on August 2, 2026.
Read definition - Compliance
GDPR
The General Data Protection Regulation (GDPR) is the European Union's data protection law, enforceable since May 2018. It governs how organizations collect, process, store, and transfer personal data of EU residents, with requirements around lawful basis, data minimization, purpose limitation, and individual rights.
Read definition - Compliance
HIPAA
The Health Insurance Portability and Accountability Act (HIPAA) is a US federal law that establishes standards for protecting sensitive patient health information (PHI). It requires covered entities and their business associates to implement administrative, physical, and technical safeguards to ensure the confidentiality, integrity, and availability of electronic PHI.
Read definition - Process
Human-in-the-Loop (HITL)
Human-in-the-loop is a design pattern in which a human reviews, approves, or overrides an AI system's decision before it is executed. In agent governance, HITL is the mechanism that ensures high-risk actions receive human judgment rather than being auto-approved by policy.
Read definition - Concept
IGA (Identity Governance and Administration)
Identity governance and administration (IGA) is the discipline of managing who has access to what, why they have it, and whether it is still appropriate. It includes user provisioning, access reviews, policy enforcement, and audit reporting across an organization's application portfolio.
Read definition - Compliance
ISO 27001
ISO/IEC 27001 is the international standard for information security management systems (ISMS). It provides a systematic approach to managing sensitive company information through risk assessment, security controls, and continuous improvement. Certification demonstrates that an organization meets globally recognized information security benchmarks.
Read definition - Compliance
ISO 42001
ISO/IEC 42001 is the first international standard for AI management systems (AIMS). Published in December 2023, it provides requirements for establishing, implementing, maintaining, and continually improving how organizations govern AI systems across their lifecycle, from design through deployment and retirement.
Read definition - Architecture
Identity Provider (IdP)
An identity provider (IdP) is a system that manages user authentication and identity data. It verifies who a user is (authentication) and, in conjunction with authorization systems, determines what they can access. Common IdPs include Okta, Microsoft Entra ID (Azure AD), and Google Workspace.
Read definition - Process
Incident Response
Incident response is the organized approach to detecting, containing, eradicating, and recovering from security incidents. An incident response plan defines roles, communication procedures, evidence preservation requirements, and escalation paths for different types of security events.
Read definition - Threat
Intent Drift
Intent drift is the gradual deviation of an AI agent's behavior from the original human objective. It can be caused by adversarial manipulation (prompt injection), probabilistic hallucination, context window degradation, or accumulated errors across a long chain of actions. Agents do not crash; they drift.
Read definition - Framework
KYA (Know Your Agent)
Know Your Agent (KYA) is a governance framework for AI agents in regulated industries, modeled on the financial sector's Know Your Customer (KYC) requirements. It defines four pillars: agent identity and registration, authority and permission control, behavior monitoring, and ecosystem governance.
Read definition - Security
Least Privilege
The principle of least privilege requires that every identity (human or machine) operates with only the minimum permissions necessary to complete its current task. No more access than needed, no longer than needed. It is a foundational concept in access management and a prerequisite for meaningful security posture.
Read definition - Protocol
MCP (Model Context Protocol)
The Model Context Protocol (MCP) is an open standard, originally released by Anthropic in late 2024, that defines how AI agents connect to external tools and data sources. It uses JSON-RPC 2.0 over standardized transports so that any MCP-compatible agent can discover and invoke any MCP-compatible server without custom integration code.
Read definition - Architecture
MCP Gateway
An MCP gateway is a reverse proxy purpose-built for AI agent traffic. It sits between AI agents (MCP clients) and the tools they access (MCP servers), intercepting every tool call to enforce authentication, authorization, and policy before the request reaches the downstream app. It is the control plane for agent-to-tool interactions.
Read definition - Security
MFA (Multi-Factor Authentication)
Multi-factor authentication (MFA) requires users to present two or more verification factors to gain access to a resource. Factors typically combine something the user knows (password), something they have (phone, hardware key), and something they are (biometric). MFA significantly reduces the risk of unauthorized access from compromised credentials.
Read definition - Compliance
NIS2
The NIS2 Directive (Network and Information Security Directive 2) is the EU's updated cybersecurity legislation for essential and important entities across critical sectors. It expands scope, strengthens security requirements, and introduces stricter incident reporting obligations and supply chain security mandates compared to the original NIS Directive.
Read definition - Concept
Non-Human Identity (NHI)
A non-human identity (NHI) is any digital identity that does not belong to a person. Service accounts, API keys, OAuth tokens, bots, and AI agents all fall under this umbrella. NHIs authenticate to systems and access data programmatically, typically without a human present at the time of access.
Read definition - Protocol
OAuth 2.0
OAuth 2.0 is the industry-standard authorization framework that enables applications to obtain limited access to user accounts on third-party services. It works by delegating user authentication to the service that hosts the account and authorizing third-party applications to access that account on the user's behalf.
Read definition - Architecture
OPA / Rego
Open Policy Agent (OPA) is a CNCF-graduated policy engine that decouples policy decision-making from application logic. Rego is OPA's purpose-built policy language: a declarative language where policies are boolean rules evaluated against structured JSON inputs.
Read definition - Compliance
PCI DSS
The Payment Card Industry Data Security Standard (PCI DSS) is a set of security requirements for organizations that handle credit card data. Maintained by the PCI Security Standards Council, it specifies controls for network security, access control, monitoring, vulnerability management, and information security policy.
Read definition - Architecture
Policy Engine
A policy engine is a system that evaluates access requests against defined rules and produces authorization decisions. It decouples policy logic from application code, allowing organizations to define, version, test, and audit their access rules independently of the systems that enforce them.
Read definition - Threat
Prompt Injection
Prompt injection is an attack in which adversarial input manipulates an AI agent into performing unintended actions. The attacker embeds instructions in data the agent processes (a document, email, web page, or tool response) that override or subvert the agent's original instructions.
Read definition - Security
RBAC (Role-Based Access Control)
Role-based access control (RBAC) is an access management model in which permissions are assigned to roles, and users are assigned to roles based on their job function. A user's access is determined entirely by the roles they hold, making it straightforward to manage permissions at scale.
Read definition - Protocol
SCIM (System for Cross-domain Identity Management)
SCIM is an open standard protocol for automating the exchange of user identity information between identity domains and IT systems. It enables organizations to automatically create, update, and deactivate user accounts across SaaS applications when employees join, change roles, or leave.
Read definition - Compliance
SOC 2
SOC 2 is an auditing framework developed by the American Institute of CPAs (AICPA) that evaluates an organization's information systems on five trust service criteria: security, availability, processing integrity, confidentiality, and privacy. A SOC 2 report provides assurance to customers that a service organization manages data with appropriate controls.
Read definition - Security
SSO (Single Sign-On)
Single sign-on (SSO) is an authentication mechanism that allows users to log in once with a single set of credentials and access multiple applications without re-authenticating. SSO reduces password fatigue, improves security by centralizing authentication, and simplifies identity management.
Read definition - Security
Service Account
A service account is a non-human identity used by applications, services, or automated processes to authenticate and interact with other systems. Unlike user accounts, service accounts are not tied to individual people and typically use API keys, certificates, or OAuth client credentials for authentication.
Read definition - Threat
Shadow AI
Shadow AI is the use of AI tools and services by employees without the knowledge, approval, or governance of IT and security teams. It is the AI-era extension of shadow IT: unauthorized technology adoption that creates unmanaged risk.
Read definition - Threat
Shadow IT
Shadow IT is the use of information technology systems, devices, software, applications, and services without explicit approval from the IT department. It includes unauthorized SaaS subscriptions, personal devices connected to corporate networks, and unsanctioned cloud services.
Read definition - Threat
Shadow MCP
Shadow MCP refers to unauthorized or unmanaged MCP server instances running inside an organization's environment without the knowledge or approval of IT or security teams. It is the MCP-layer extension of shadow IT: employees installing and connecting AI agent tools outside of centralized governance.
Read definition - Compliance
Sub-Processor
A sub-processor is a third party engaged by a data processor to process personal data on behalf of the data controller. Under GDPR, processors must obtain prior written authorization from the controller before engaging sub-processors, and must ensure sub-processors meet the same data protection obligations.
Read definition - Security
Supply Chain Security
Supply chain security is the practice of managing risks introduced by third-party vendors, software dependencies, and service providers that an organization relies on. In AI agent contexts, it extends to MCP servers, agent platforms, model providers, and the tools agents connect to.
Read definition - Concept
Suspend and Resume
Suspend and resume is a governance capability in which an AI agent pauses mid-task when it encounters an action that requires human approval, waits for a policy decision, and then continues with full context once the decision resolves. The agent experiences the pause as a slow tool call, preserving its working state.
Read definition - Concept
TBAC (Task-Based Access Control)
Task-based access control (TBAC) is an authorization model in which the task, not the role or attribute, is the unit of authorization. Permissions are synthesized just-in-time for each task and revoked on completion, scoping access to the lifecycle of a specific piece of work.
Read definition - Protocol
XAA (Cross App Access)
Cross App Access (XAA) is an open protocol backed by Okta that extends OAuth to secure agent-driven and app-to-app interactions across the enterprise. It shifts authorization control from individual applications to the identity provider, enabling centralized visibility and policy enforcement over how AI agents connect to downstream apps.
Read definition - Concept
Zero Standing Permissions
Zero standing permissions (ZSP) is a security model in which no user, agent, or system retains persistent privileged access. Elevated permissions are granted only when needed, scoped to a specific task, and revoked on completion. In the context of AI agents, ZSP means an agent starts every session with no access and receives only what the current task requires.
Read definition - Security
Zero Trust
Zero trust is a security architecture that requires every access request to be verified, regardless of where it originates or what resource it accesses. It operates on the principle of "never trust, always verify": no user, device, or system is trusted by default, even if it is inside the corporate network perimeter.
Read definition