Your assistant must block a list of topics and mask personal data in replies, without changing each app. What do you use?
Why D is the answer
Amazon Bedrock Guardrails
Guardrails apply these policies consistently and independently of the model. Prompt instructions are easy to bypass.
In detail
- Guardrails apply central policies, with no change to each app.
- Guardrails work independently of the model.
Use case
An illustrative scenario, not a real incident.
Five apps need the same topic blocks and masking. A shared guardrail applies them all, with no change to each app.
See it in different ways
flowchart LR A["User input"] --> B["AWS WAF"] B --> C["Input guardrail"] C --> D["Foundation model"] D --> E["Output guardrail"] E --> F["Grounding check"] F --> G["Response"]
sequenceDiagram participant U as User participant I as IAM participant G as Guardrails participant M as Model participant L as CloudTrail U->>I: Is this caller allowed? I->>G: Filter the input G->>M: Safe prompt M-->>G: Output G-->>U: Filtered output I->>L: Records the call
mindmap
root(("Bedrock Guardrails"))
n1["Topics"]
n2["Denied topics"]
n3["Privacy"]
n4["Mask personal data"]Key point: Guardrails apply central content policy.
Why the other options are wrong
- A. An S3 bucket policy
- A bucket policy controls S3.
- B. A CloudWatch alarm
- An alarm reports metrics.
- C. Instructions in each system prompt
- Prompts must change in each app, and they do not guarantee results.