AISP practice · Module 5

Secure AI Operations and Incident Response — 15 questions. Choose the BEST answer.

You can answer as a guest, but you'll need a free account to see your score and save it. Sign up · Sign in
Question 1 Question 1 · M5
Investigators need to know whether a RAG assistant's permission check ran for a request that leaked another client's data. Which log is essential?
Question 2 Question 2 · M5
Where should an AI kill switch operate?
Question 3 Question 3 · M5
Human reviewers of an AI claims model normally override 6% of decisions. This week the rate is 0.2%. How should operations treat this?
Question 4 Question 4 · M5
Why should legal and privacy join an AI incident early?
Question 5 Question 5 · M5
A marketing manager edited a production system prompt through an admin console on a weekend, and the assistant began quoting wrong rates. Which control would MOST directly have prevented this?
Question 6 Question 6 · M5
Poisoned data was used to fine-tune a model, which was then distilled into two smaller models. What must eradication include?
Question 7 Question 7 · M5
After a poisoning incident, what is the safest way to recover a vector index?
Question 8 Question 8 · M5
A guardrail fix for a jailbreak fails once when the tester retries the original prompt. Can the ticket be closed?
Question 9 Question 9 · M5
A unique string planted in a chatbot's system prompt appears in a post on a public forum. What does this MOST likely indicate?
Question 10 Question 10 · M5
Which AI weakness should usually be prioritized FIRST?
Question 11 Question 11 · M5
A compromised agent has queued outbound payments. When stopping it, what should happen FIRST?
Question 12 Question 12 · M5
Injected instructions arrive through one poisoned knowledge-base article. What is the BEST first containment step?
Question 13 Question 13 · M5
An API key appears in a prompt that a developer pasted into an internal assistant. What should happen FIRST?
Question 14 Question 14 · M5
The SOC wants to hunt for prompt injection, but the privacy officer objects to analysts reading raw customer prompts. What design satisfies both?
Question 15 Question 15 · M5
A summarization agent starts calling a delete tool it has never used, seconds after reading emails from one external sender. What is this MOST likely?