Questions
Common questions.
What is Prompt Injection?
An attack where malicious instructions are embedded in data that an AI agent processes, causing it to deviate from its intended behavior.
How is Prompt Injection used in production?
This can lead to unauthorized data access, tool misuse, or policy bypasses.
Which terms are related to Prompt Injection?
Closely related concepts include Agent Hijacking, Trust Chain, Adversarial Attack, Agent Retirement. Each is defined in the Prefactor glossary.
Keep reading