StudyByteLearn
← SC-100: Microsoft Cybersecurity Architect
SC-100 · SaaS PaaS IaaS containers and AI services

Practice question 340 of 498 · hard

An AI agent can propose transferring confidential data to an external service. What design best limits high-risk action impact?

Choose the single best answer.

  1. A.Rely on output formatting to prove the destination is approved
  2. B.Use a private endpoint as automatic authorization for the transfer
  3. C.Execute every proposal because a system prompt says the agent is safe
  4. D.Require an independent approved workflow and scoped action permissions
Show the answer and explanation

✓ Correct answer

  • Require an independent approved workflow and scoped action permissions

Human approval for high-risk operations and scoped permissions control authority independently of the model. System prompts guide behavior but do not confer safety guarantees. Private networking and output format do not establish transfer authorization.

Read the source reference

Read and review here, or open the app to practise and save your results on this device.

Practise in the app

Unofficial study material. Not affiliated with or endorsed by Microsoft. These are original learning questions, not actual exam questions. Product capabilities and exam objectives can change; verify current Microsoft documentation. Practice scores do not predict or guarantee an exam result.