AI Glossary
Prompt injection
What does “Prompt injection” mean?
Prompt injection: Attack in which text the agent reads, such as a web page, email or document, contains instructions meant to hijack it. Because models process instructions and data as the same kind of text, this remains one of the main unsolved security problems for agents.
What is an example of “Prompt injection”?
A web page could hide text telling a browser agent to send the user's files elsewhere; well-designed agents ignore such instructions.