Skip to content
AI Security

Interview question · How LLM Systems Work

Beyond prompt injection, what security problems come with accepting images, audio and documents?

Uploads bring classic file-handling bugs into AI products. The scored answer covers the parsers, fetchers and storage around the model, not only what the model reads.

mediumNewUpdated Nov 2026

Uploads bring classic file-handling bugs into AI products. The scored answer covers the parsers, fetchers and storage around the model, not only what the model reads.

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 5 follow-ups like an interviewer would

An original fictional museum lets visitors submit photos and scanned catalog pages. Staff can also process restricted conservation scans. A worker produces previews and extracted text before calling a model.

  1. What if an upload is small on disk but expands into a large image?
  2. What if the text extraction service is approved for public catalog scans only?
  3. What if a permitted URL redirects to a destination outside the fetch policy?
  4. What if deleting the upload leaves its preview visible through a public URL?
  5. What is the release evidence?

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 changes for security when an application uses a reasoning model that thinks before it answers?Next in this trackDescribe a RAG pipeline as a data flow and mark every point an attacker can influence.Next in this trackHow does function calling work, and where does an agent actually get its authority?
DISCUSSION · 0

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