Question 70
At Aster Retail, a storefront service desk is improving order-support before expanding the internal rollout (SF-P20-02). The administrator observes that a retrieved knowledge article includes instructions telling the assistant to ignore the company policy. Any fix must ensure that customer order details stay within each authorized account. Which underlying Agentforce or Salesforce design principle explains the appropriate approach?
Explanation
Correct Answer
The governing distinction is: External content must not become trusted operational instruction merely because it is retrieved. In this order-support scenario, this separates supported agent behavior from assumptions about the AI model. Because a retrieved knowledge article includes instructions telling the assistant to ignore the company policy, the intended response is to treat retrieved text as data, enforce higher-priority agent instructions and permission controls. An observable check is that a malicious-article test does not trigger unauthorized actions or policy overrides.
Incorrect Answers
The idea “The Trust Layer supplies security and privacy safeguards but cannot replace data permissions; interpret this while preserving the specified source of truth” explains another source-defined situation: A case-summary prompt may contain personal contact information that must not be exposed beyond authorization. The correction must ensure that student records never cross program authorization boundaries. That situation depends on: The Trust Layer supplies security and privacy safeguards but cannot replace data permissions. The present question instead concerns a retrieved knowledge article includes instructions telling the assistant to ignore the company policy. The correction must ensure that customer order details stay within each authorized account. Its directly relevant mechanism is: External content must not become trusted operational instruction merely because it is retrieved.
The idea “Trust safeguards complement rather than supersede object and field authorization; interpret this as distinct from unrelated UI configuration” explains another source-defined situation: A manager assumes Trust Layer usage automatically grants agents unrestricted access to customer records. The correction must ensure that employee identity is verified before accessing private tickets. That situation depends on: Trust safeguards complement rather than supersede object and field authorization. The present question instead concerns a retrieved knowledge article includes instructions telling the assistant to ignore the company policy. The correction must ensure that customer order details stay within each authorized account. Its directly relevant mechanism is: External content must not become trusted operational instruction merely because it is retrieved.
The idea “Prompt Builder generates contextual text and should not replace deterministic business-rule execution; interpret this without inferring extra permissions from an LLM response” explains another source-defined situation: A nightly data cleanup requirement is being treated as a request for open-ended generated prose. The correction must ensure that customer order details stay within each authorized account. That situation depends on: Prompt Builder generates contextual text and should not replace deterministic business-rule execution. The present question instead concerns a retrieved knowledge article includes instructions telling the assistant to ignore the company policy. The correction must ensure that customer order details stay within each authorized account. Its directly relevant mechanism is: External content must not become trusted operational instruction merely because it is retrieved.