Prompt Injection

AI Governance

A security vulnerability where an attacker injects malicious instructions into a prompt to make an AI model behave unexpectedly. This can lead to data leaks, policy violations, or other harmful outcomes.

← Back to glossary