Skip to content
AI Security

Interview question · Cloud, Container & AI Infrastructure Security

How do you protect a public AI API from denial of service and abuse at the edge?

An AI request can cost a thousand times more than a web request, so volumetric thinking alone fails. The scored answer layers network, request and cost controls, and protects the GPU tier from ever seeing junk.

mediumNewPreparation tags: Amazon, Google, MicrosoftUpdated Nov 2026

An AI request can cost a thousand times more than a web request, so volumetric thinking alone fails. The scored answer layers network, request and cost controls, and protects the GPU tier from ever seeing junk.

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 6 follow-ups like an interviewer would

An AI API is hit by a flood of expensive requests that drives up GPU costs and slows service for paying customers.

  1. What absorbs volumetric attacks?
  2. How do you protect at the request layer?
  3. Why meter tokens rather than requests?
  4. How do you detect abuse?
  5. How do you keep paying customers served?
  6. What layered defence would you build?

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 trackHow do you manage insider risk for model weights and research IP?Next in this trackExplain RAND's security levels for model weights and which attacker each is meant to stop.Next in this trackDesign backup, rollback and recovery for an AI system's model, prompts, retrieval index and tool configuration.
DISCUSSION · 0

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