Question 390
At fictional Meadowford Care Network, a Snowflake data and AI engineering team is conducting an accountable generative AI operations change under an approved model lifecycle and release-control process. The separately composed technical incident is: A platform team adds detailed tracing to troubleshoot an agent, but the default trace captures unredacted customer prompts and model outputs in a broadly shared dashboard. Assume only these documented facts and use safe, authorized production-change procedures; exact current Snowflake feature syntax and region or account availability require current review. Which repeatable assurance or control is most relevant to preventing this situation?
Explanation
Writer-proposed answer: B (UNVERIFIED)
Preliminary individual option explanations — NOT expert validated
Correct Answer
Review tracing fields, apply role-based access and retention controls and run periodic log-sensitivity audits. Observability must be designed with privacy boundaries, not treated as permission to duplicate all data in logs.
Incorrect Answers
This response concerns Gen AI runaway cost and chargeback, which is different from the stated failure. A platform team adds detailed tracing to troubleshoot an agent, but the default trace captures unredacted customer prompts and model outputs in a broadly shared dashboard.
Agent tool execution requiring approvals is not the cause of the reported failure: A platform team adds detailed tracing to troubleshoot an agent, but the default trace captures unredacted customer prompts and model outputs in a broadly shared dashboard.
Cortex model access allowlist governance alone does not establish the needed evidence: Debugging remains possible from safe trace identifiers and performance fields while access tests cannot reveal restricted prompt payloads.