Cybersecurity Glossary

Stored context an agent can retrieve across steps or sessions, requiring integrity and access controls.

An AI-enabled system that plans and invokes tools to pursue a goal with some degree of autonomy.

A record of where AI-related data came from, how it changed, and where it was used.

A structured review of an AI use case, assets, harms, threats, controls, and residual risk.

The governance and technical discipline of protecting AI systems, their data, users, and connected services.

The governed use of AI to assist security monitoring, triage, investigation, and response workflows.

Adding trusted context such as identity, asset, and threat information to a security alert.

An API flaw where a caller can access another object's data by changing an identifier.

The disciplined design, testing, deployment, and maintenance of security detections.

An open protocol through which AI applications connect to external data sources, tools, and workflows.

Instructions in user or retrieved content that attempt to redirect an AI system from its intended policy.

Manipulation of tool metadata or behavior so an AI system is induced to take an unsafe action.