Prompt injection
Prompt injection is an attempt to make an AI system follow instructions inserted into content that it should treat as data
What it means, how it is used and where it comes up, with its related concepts.
What it is and how it works
It can appear in external documents, pages or messages. Protection requires separating trust levels, limiting permissions and evaluating actions. Asking the model to ignore malicious instructions is not in itself a sufficient defence.
Topics