NCA-GENL Trustworthy AI Practice Question
A retail company is using NVIDIA NeMo Guardrails to build a customer-facing shopping assistant. The security team wants to prevent users from extracting the system prompt or instructing the model to ignore its safety rules. Which guardrail type should be configured first to intercept these attempts before they reach the LLM?
⚠ Common exam trap
The trap here is choosing output rails because they can detect leaked content, when the requirement is to stop the malicious instruction before the model processes it.
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
✓
Input rails that detect and block prompt injection and jailbreak patterns in the user message.
Input rails inspect the user message before it reaches the model, making them the correct first layer to block prompt injection and jailbreak attempts. By rejecting malicious inputs early, the assistant never processes instructions that could extract the system prompt or bypass safety rules. Output and retrieval rails address different stages and are complementary, not primary, for this threat.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Dialog rails that redirect the conversation to a fallback topic when the user asks about shopping.
Why it's wrong here
Dialog rails manage conversational flow and intent routing, but a fallback topic triggered by shopping questions does not address malicious prompt injection. The scenario requires blocking adversarial inputs, not redirecting legitimate shopping conversations. This misapplies the rail type and would degrade the user experience for normal customers.
- ✗
Retrieval rails that filter the documents returned by the RAG pipeline before they are added to the context.
Why it's wrong here
Retrieval rails govern which retrieved documents enter the context window, protecting against poisoned or irrelevant knowledge. They do not inspect the user's message for injection attempts, so they cannot block a user trying to extract the system prompt or override safety instructions. This is the wrong layer for the stated threat.
- ✗
Output rails that scan the model's response for sensitive content before returning it to the user.
Why it's wrong here
Output rails act after the model has generated a response, so they can catch leaked content but do not prevent the model from processing a prompt injection in the first place. For prompt extraction and jailbreak attempts, intercepting at the input stage is more effective. Output rails are a valuable complementary layer but not the first line of defense for this scenario.
- ✓
Input rails that detect and block prompt injection and jailbreak patterns in the user message.
Why this is correct
Input rails evaluate the user message before it reaches the LLM, so they can block prompt injection and jailbreak attempts at the earliest point. This prevents the model from ever processing a malicious instruction, which is exactly what the security team wants for prompt extraction and safety-rule bypass attempts. It is the correct first layer for this threat.
About these practice questions
Courseiva writes every NCA-GENL question from scratch — 367 in total, each with an explanation and a wrong-answer breakdown. None are copied from real exams or dumps. 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 NVIDIA exam blueprint
This NCA-GENL practice question is part of Courseiva's free NVIDIA 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 NCA-GENL exam.