An AI may have returned confidential information.
Investigate impact and prevent recurrence.
Choose two first-response actions.
Preserve model, prompt, retrieved-document, authorization, and timestamp evidence, contain exposure, and start the reporting process.
Detailed explanation
The event can be reproduced and scoped.
The event can be reproduced and scoped.
Containment and organizational response proceed together.
Containment and organizational response proceed together.
Cause and impact cannot be established.
Cause and impact cannot be established.
Incorrect information can create further harm.
Incorrect information can create further harm.
Try it yourself
An example you can run in a temporary verification environment.
AWS公式AIF-C01 Domain 5.2のインシデント対応、監査、データ保持を確認する。Expected result
AI出力の事故で証跡を保全し、被害抑制と報告を進める初動を説明できる。Key points
- Evidence preservation
- Containment
- Reporting
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
- Command output formatting can vary slightly by distribution or tool version.
- Run the example in a temporary directory or process when possible.
Foundation review
Read the scope first
Check whether the command acts on the current shell, a new process, an existing process, or a file.
Verify the observable result
Use the supplied command and compare the output with the expected result.