PLAT-ADMN-201 · Question #88
PLAT-ADMN-201 Question #88: Real Exam Question with Answer & Explanation
The correct answer is B: Ensure the users' permissions and field-level security restrict access to the CEO's health plan.. The most secure and reliable way to prevent an AI agent from exposing sensitive data is to restrict access to that data at the platform level using Salesforce's built-in security model - object permissions and field-level security (FLS). If the underlying data is inaccessible to
Question
Cloud Kicks has implemented an Employee Agent to answer benefits questions for its employees. How should a Platform Administrator prevent the agent from responding to staff members' questions about the CEO's private health plan and benefits?
Options
- AConfigure assignment rules to assign the agent to employee data.
- BEnsure the users' permissions and field-level security restrict access to the CEO's health plan.
- CModify the agent's instructions and guardrails to block questions related to the CEO's health plan.
- DTrain the agent on employee health plans instead of the CEO's health plan.
Explanation
The most secure and reliable way to prevent an AI agent from exposing sensitive data is to restrict access to that data at the platform level using Salesforce's built-in security model - object permissions and field-level security (FLS). If the underlying data is inaccessible to the agent's running context, it physically cannot retrieve or surface that information, regardless of how the agent is prompted. Option C (guardrails/instructions) is a soft control that can be bypassed or circumvented by clever prompting. Option D (training) doesn't prevent access to existing data. Option A (assignment rules) controls case routing, not data visibility.
Topics
Community Discussion
No community discussion yet for this question.