QuestionQ61

VerifiedAI Safety, Security, and Governance

A bank is building a generative AI (GenAI)-powered AI assistant that uses Amazon Bedrock to help the bank’s website users with account inquiries and financial guidance. The bank must ensure that the AI assistant does not disclose any personally identifiable information (PII) during customer interactions.

The AI assistant must not transmit PII in prompts to the GenAI model. It must not respond to customer requests for investment advice. The bank must gather audit logs for all customer interactions, including any images or documents transmitted during those interactions.

Which solution meets these requirements with the LEAST operational effort?

Explanation

Amazon Bedrock Guardrails can apply sensitive-information filters to input prompts and model responses, blocking or masking PII, and can use a denied-topics policy to block investment-advice content. Amazon Bedrock model invocation logging supports Converse API calls; when Amazon S3 delivery and image logging are enabled, images and documents passed through Converse are logged to Amazon S3. These native managed controls satisfy the safety and audit requirements without custom detection or prompt-engineering logic.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!