Skip to content
AI Security

Interview question · Prompt Injection & Jailbreaks

How would you test whether a RAG assistant is vulnerable to indirect prompt injection, without harming production?

Trace editable source material through retrieval, actual context, model behavior and observed effects. Work through a synthetic result with denied proposals and an unresolved receipt.

mediumNewUpdated Nov 2026

Trace editable source material through retrieval, actual context, model behavior and observed effects. Work through a synthetic result with denied proposals and an unresolved receipt.

Make your first attempt

What would you say before reading the answer?

  1. 01 State the assumptions your answer depends on.
  2. 02 Explain the mechanism and choose a control.
  3. 03 Name a condition that would change your decision.

The worked answer includes an approach, scoring criteria and follow-up questions. Compare the reasoning with your own.

10 answers per topic instead of 5, plus saved progress and bookmarks · costs nothingor unlock all 529 remaining answers · $25

Every worked answer in all 17 topics, every lesson in every course, the full concept curriculum, the complete practice-test bank, and the downloadable security field guides.

6 months · One payment · No auto-renewal

What if · the follow-ups in this answer

The answer works through 4 follow-ups like an interviewer would

Marc evaluates a fictional procurement assistant with synthetic suppliers and buyer grants. A supplier controls its description, while the test harness records context assembly, update decisions and controlled effects.

  1. The controlled passage appears in search results but never reaches the model. What passed?
  2. Four unauthorized proposals occur, three are denied and one has no effect receipt. How do you report it?
  3. A citation names another document. Does that establish unauthorized access?
  4. What evidence would change your release recommendation?

Each rung has a nudge, a model answer, the principle that carries over and the weak answer to avoid. Premium opens every answer and ladder.

UP NEXT ON YOUR JOURNEY
Next in this trackWhat evidence should decide an AI assistant's residual-risk target?Next in this trackMap a multi-stage indirect prompt injection onto a kill chain. Where would you place detections and controls at each stage?Next in this trackHow do attackers steal what is in an LLM application's context: system prompts, retrieved documents, other users' data and tool results?
DISCUSSION · 0

No comments yet. Share a correction or a different way to answer it.