AI Glossary

Indirect prompt injection

What does “Indirect prompt injection” mean?

Indirect prompt injection: Instructions hidden in outside content that an AI system reads, such as a document or webpage.

Indirect prompt injection visual guide: Instructions hidden in outside content that an AI system reads, such as a document or webpage.
Visual guide Open full size (new tab)

What is an example of “Indirect prompt injection”?

A webpage tells an agent to ignore the user's task and send information elsewhere.

Keep learning

See also

AlignmentTermAI safetyTermAgent guardrailsTermAdversarial exampleTermAdversarial trainingTerm

Same topic

Data exfiltrationTermLeast privilegeTermJailbreakTermHumanEvalTermLeaderboardTerm