A compliance officer asks whether prompts sent to Amazon Bedrock are used to train foundation models. What's the documented position?
A.Prompts are used to improve models unless you opt out
No opt-out needed — training use is off by default.
B.Customer prompts and completions are NEVER used to train or improve the base models
Documented no-training commitment. Correct.
C.Only anonymized prompts are used
Not even anonymized data is used.
D.Prompts train models but only in your region
Data doesn't train models — regionality is about residency, not training.
Why this answer
Bedrock's data-protection commitment: prompts and completions are NOT used to train or improve the base FMs, aren't shared with model providers, and stay within your region — a frequently tested AIF governance fact.
Correct answer: B · Last reviewed 2026-11-05
Official AWS references
Practice this interactively
Answer without seeing the solution first in the AI Security & Governance practice session.
Practice AI Security & Governance