AI security glossary

AI Security Glossary.

Plain-language definitions for the AI security terms that matter. Linked, citable, and short.

By the numbers
31
Terms
5
Frameworks
26
Concepts
All terms

Every term, A to Z.

Jump to a letter above, or scan the full list below.

Adversarial Testing (AI)

Systematic testing of an AI system against attacks, edge cases, and failure modes.

Read the definition

Agentic AI

AI systems that take actions in the world via tool use rather than only producing text. Includes custom GPTs, autonomous agents, and tool-using LLM applications.

Read the definition

AI Firewall

Network-layer security control that inspects AI traffic (prompts, outputs, API calls) for policy violations, sensitive data, threats.

Read the definition

AI Governance

The organizational structures, accountabilities, and processes for managing AI risk and ensuring responsible AI use.

Read the definition

AI Policy (Internal)

Organizational policy governing AI use: what's allowed, what's prohibited, who approves, how violations are handled.

Read the definition

AI Red-Teaming

Adversarial testing of AI systems to identify safety, security, and robustness failures before production.

Read the definition

Constitutional AI

Anthropic's approach to training language models with a defined set of principles (a constitution) used during fine-tuning to bias toward safer outputs.

Read the definition

Custom GPT

OpenAI's brand for user- or organization-built variants of ChatGPT with custom instructions, knowledge, and tool access.

Read the definition

Differential Privacy

A mathematical framework for measuring and bounding the privacy loss when statistics or models are released from sensitive data.

Read the definition

Embedding Inversion

Recovering source text or sensitive features from vector embeddings without direct access to the original content.

Read the definition

EU AI Act

Regulation (EU) 2024/1689. Risk-tiered AI obligations including prohibited practices, high-risk system requirements, transparency obligations.

Read the definition

Excessive Agency (LLM06)

An LLM-based agent has more permissions, tool access, or autonomy than its task requires.

Read the definition

Hallucination (LLM)

When an LLM generates plausible but factually incorrect content.

Read the definition

Indirect Prompt Injection

Prompt injection delivered through retrieved or referenced content (web pages, documents, emails) rather than direct user input.

Read the definition

ISO/IEC 42001

International standard for AI management systems. Certifiable. Published 2023.

Read the definition

Jailbreak (LLM)

A specific class of prompt injection that bypasses an LLM's safety training to elicit content the model was tuned to refuse.

Read the definition

LLM Firewall

A security control that inspects prompts and outputs of LLM applications for policy violations, prompt injection, sensitive data, or jailbreak patterns.

Read the definition

MITRE ATLAS

Adversarial Threat Landscape for AI Systems. MITRE's catalog of adversarial machine learning tactics and techniques. Modeled on MITRE ATT&CK.

Read the definition

Model Card

Standardized documentation of an AI model's intended use, training data, performance, limitations, and ethical considerations.

Read the definition

Model Inversion

An attack that recovers training data or sensitive features by querying a model and analyzing outputs.

Read the definition

NIST AI RMF

NIST AI Risk Management Framework. Voluntary US framework defining four functions (Govern, Map, Measure, Manage) for AI risk.

Read the definition

OWASP LLM Top 10

OWASP's catalog of the top 10 risks for LLM applications. Updated annually. The most-cited LLM security framework.

Read the definition

Prompt Injection

An attack where crafted input causes an LLM to override its instructions or context. Direct injection comes through user input. Indirect injection comes through retrieved or referenced content the LLM processes.

Read the definition

RAG Security

Security considerations specific to Retrieval-Augmented Generation pipelines: vector-store access control, corpus integrity, embedding inversion, indirect prompt injection.

Read the definition

RLHF (Reinforcement Learning from Human Feedback)

Technique for fine-tuning language models using human feedback to align outputs with preferred behaviors.

Read the definition

Shadow AI

Unsanctioned AI use within an organization, including consumer-tier ChatGPT, Copilot trial, custom GPTs, browser extensions.

Read the definition

Sycophancy (LLM)

An LLM's tendency to agree with the user's stated position or assumption rather than provide accurate analysis.

Read the definition

System Card

OpenAI's term for documentation of an AI system's safety properties, limitations, and risk evaluations.

Read the definition

Training Data Poisoning

Adversarial manipulation of training, fine-tuning, or RAG-corpus data to alter model behavior.

Read the definition

Vector Store

A database optimized for similarity search over high-dimensional embeddings. Foundational component of RAG pipelines.

Read the definition

Vendor Due Diligence (AI)

The process of assessing an AI vendor's security, privacy, and operational posture before procurement.

Read the definition
AI Posture Check

Want a personalized read?

Take the free AI Posture Check. The glossary terms become specific recommendations for your environment.