Skip to main content

AI Security 
Glossary

Get familiar with key terms and concepts in the AI Security space.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

Prompt Injection

Prompt injection is an attack where crafted input causes a large language model to deviate from its intended instructions and follow the attacker's instead.

Prompt Leak

Prompt leak is a form of prompt injection where a model is manipulated into revealing its system prompt or internal instructions, exposing what is effectively proprietary configuration.

Shadow AI

Shadow AI describes the AI tools and agents employees adopt on their own, from chat assistants to coding copilots to autonomous agents, without visibility or approval from IT or security teams.

Toxic, Biased or Harmful AI-generated Content

An LLM, jailbroken or not, can produce toxic, biased, or otherwise harmful output. Because model behavior is non-deterministic, this risk exists even when a system is working exactly as designed.

Visual Prompt Injection

In a visual prompt injection, malicious instructions are hidden inside an image rather than the text prompt, designed to be read by the model but invisible to the person looking at it.