Skip to content
AI Security
Practice tests · 50 items

AI Red Teaming, Evaluation & Response practice test

Scoping an engagement from harms, reconnaissance for AI applications, manual and automated testing, scoring and LLM-as-judge reliability, attack success as a rate with confidence intervals, reporting, regression suites, monitoring, AI incident response and forensics, and using AI in security operations. The pool for this topic holds 12 easy, 22 medium and 16 hard items. Each answer comes with its reasoning and a link to the question or concept page it was drawn from.

Configure a test
Topic
Difficulty
Questions
12 in this pool · about 7 min
Reveal answers
Sign in to startFree account required · retakes favour new items

Sample items with answers

easy · sample
A prior penetration test included the assistant’s retrieval identity and draft readers. What should an AI red-team reviewer do with that work?
Repeat the entire test because penetration testing cannot assess model-mediated behavior.
✓Review its scope and reuse applicable evidence; assess changed or uncovered paths.
Accept it as complete assurance because the assistant’s permission checks were mentioned.
Exclude its findings because AI red teaming only evaluates a model’s generated text.

Coverage comes from the executed protocol, configuration and observations. Penetration testing can include AI workflows, while a red-team exercise can include conventional permission checks. Reuse evidence that still matches the requirement and add work for unresolved or changed paths. Neither label establishes complete coverage.

easy · sample
An assessment reports a 4% attack success rate. Which information is needed to interpret that percentage?
✓The success event, trial unit, counts, selection, budget, configuration and scoring method.
Only the model name, because attack success has the same meaning for every application.
The number of named techniques, because each technique is always one independent trial.
Only the severity score, because the trial count does not affect the interpretation of a rate.

A percentage needs a defined numerator and denominator under a stated protocol. Report unresolved effects and execution gaps as well as successes. The observed result may describe selected cases without estimating future traffic or adaptive campaigns. Findings and allowed-task results remain necessary alongside the rate.

After the test

Choosing from options is easier than reasoning it out in front of a panel. These pages make you reason it out: