CCAR-P Governance, Safety, and Risk Management Practice Question
A financial services firm runs a Claude-powered loan pre-screening agent that reads applicant emails and drafts a preliminary recommendation. The firm's risk committee insists that no automated decision may be finalized without a documented human review step. Which control most directly satisfies this requirement while preserving the agent's throughput?
⚠ Common exam trap
The trap here is assuming that making the model more cautious, faster, or better logged substitutes for an actual human decision point before the output is finalized.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Insert a mandatory human-in-the-loop approval gate before any recommendation is written back to the loan origination system.
The risk committee's requirement is about who authorizes a decision, not how the model behaves or how well activity is recorded. Only a human-in-the-loop approval gate makes a person the final authorizing step before the recommendation reaches the loan origination system, yielding the documented review the committee expects while letting the agent handle upstream work automatically.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Log every model request and response to immutable storage and retain the logs for seven years.
Why it's wrong here
Comprehensive logging supports audit and forensics, but it is a detective control, not a preventive one. The recommendation would still be finalized automatically; logs would merely record that it happened, leaving the requirement for documented human review before finalization unsatisfied.
- ✗
Increase the model's temperature to zero and add a system prompt instructing Claude to be conservative in its recommendations.
Why it's wrong here
Deterministic sampling and a cautious system prompt change how the model reasons, but they do not insert any human judgement into the workflow. A conservative automated recommendation is still an automated decision, so the committee's requirement for documented human review before finalization remains unmet by this approach.
- ✓
Insert a mandatory human-in-the-loop approval gate before any recommendation is written back to the loan origination system.
Why this is correct
A human-in-the-loop gate places a person between the model's draft and the system of record, so an automated decision never becomes final without documented review. The agent still reads and drafts at machine speed, and only the last write is gated, which preserves throughput while producing the auditable review artefact the risk committee demands.
- ✗
Enable prompt caching on the applicant-email summarization step to reduce latency and cost per screening.
Why it's wrong here
Prompt caching is a performance and cost optimization that reuses previously processed context; it has no bearing on whether a human reviews the output. It would make the pipeline faster and cheaper, but the governance gap around unreviewed automated decisions would persist unchanged.
About these practice questions
One of 262 original CCAR-P practice questions on Courseiva, each with a full explanation and wrong-answer analysis — not exam dumps or protected exam content. Learn why practice questions differ from exam dumps →
JA
Written and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official Anthropic exam blueprint
This CCAR-P practice question is part of Courseiva's free Anthropic certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the CCAR-P exam.