Glossary

Corrupting the data an AI model trains or retrieves from so that the model learns, retrieves, or repeats attacker-chosen behavior.

Endpoint Detection and Response: an agent that records endpoint activity, detects suspicious behavior, and lets responders isolate or remediate machines remotely.

The principle that every user, process, and system gets exactly the access it needs to do its job, and nothing more.

Fraudulent messages that impersonate a trusted party to trick people into revealing credentials, paying money, or running malware.

An attack that smuggles instructions into an AI system's input so the model follows the attacker's intent instead of the operator's.

Malware that encrypts or steals data and demands payment for its return, increasingly paired with the threat of public leaks.

The amount and type of risk an organization is willing to accept in pursuit of its objectives, set by leadership and used to guide decisions.

Technology adopted by teams without the knowledge or approval of IT and security, from unsanctioned SaaS to personal devices.

Security Information and Event Management: a platform that collects and correlates logs across systems so analysts can detect and investigate incidents.

A security model that treats every request as unauthenticated and unauthorized until proven otherwise, regardless of where it comes from.