5.1Standard
A public chatbot is evaluated before launch.
Find failures that normal tests miss.
Which evaluation is most appropriate?
× Incorrect
Run an independent red-team test with injection, extraction, harmful requests, and bypass inputs to expose safety weaknesses.
Detailed explanation
IncorrectUse only the public benchmark score
It does not measure your attack surface.
IncorrectAvoid dangerous inputs as the safety control
Unverified weaknesses remain.
IncorrectRun only large numbers of normal questions
Adversarial failures remain untested.
CorrectRun an independent red-team test with injection, extraction, harmful requests, and bypass inputs
Attacker perspectives reveal safety weaknesses.
Check official documentation
Compare the following points with the official documentation.
AWS公式AIF-C01 Domain 4の安全性評価、レッドチーム、ガードレールを確認する。
Official references
What to confirm
正常系の精度だけでなく、敵対的入力に対する安全性を評価できる。
Key points
- Red team
- Adversarial input
- Safety evaluation
Notes
- Environment: AWS公式AIF-C01試験ガイドとAWS公式ドキュメントの確認
- Compare the definition with the AIF-C01 objective and the service documentation before applying it.
- Keep the concept separate from a particular answer choice or implementation step.
Foundation review
Red-team test
An authorized exercise that probes a system for security or safety weaknesses.
Evasion input
An input crafted to bypass detection, policy, or intended behavior.
Question IDAWS-AIF-543
EnvironmentAWS公式AIF-C01試験ガイドとAWS公式ドキュメントの確認
Last technical review2026-08-20
Report an error or rights issue