Skip to main content

Prompt Injection

Also known as: Prompt Attack, Jailbreaking

In one sentence

A security vulnerability where users trick an AI into ignoring its instructions by inserting malicious commands into their prompts.

Explain like I'm 12

Like convincing a guard to ignore the rules by sneaking special instructions into your conversation that make them think they should do what you say instead.

In context

Example: Adding 'Ignore all previous instructions and...' to bypass content filters or system prompts. Defended against with guardrails and input validation.

See also

Related Guides

Learn more about Prompt Injection in these guides: