Prompt Injection Prevention Quiz

What you'll be asked

1. What is the primary goal of a prompt injection attack?
2. Which of the following is an effective technique for mitigating prompt injection at the system level?
3. Why is 'sandwiching' user input between system instructions considered a partial but insufficient defense?
4. When implementing a 'dual LLM' pattern for security, what is the role of the second, smaller model?
5. A system uses a retrieval-augmented generation (RAG) pipeline. An attacker inserts instructions into a document that the RAG system retrieves. What is this specific type of attack called?

Answers and explanations come after you submit.