NCP-GENL Safety, Ethics, and Compliance Practice Question

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Enable request-level rate limiting per user account to reduce the volume of generated articles entering the editorial pipeline.

  Why it's wrong here

  Rate limiting is an abuse and cost control, not a content-safety control. It does nothing to detect verbatim reproduction of copyrighted text or defamatory assertions about individuals. Even a single generated article can infringe or defame, so throttling the request count leaves both flagged risks fully intact while adding friction for legitimate users.

• ✗

  Raise the model's temperature and top-p values so generations vary more and are less likely to match any single source.

  Why it's wrong here

  Increasing sampling randomness does not prevent memorized sequences from surfacing, because highly memorized passages tend to dominate the distribution and can still be emitted under high-temperature sampling. It also degrades output quality and makes the drafting assistant less useful. Randomness is not a copyright control and provides no protection against defamatory claims about real people.

• ✓

  Deploy a retrieval-based similarity check that compares generated output against a licensed corpus index and blocks or rewrites spans that exceed a verbatim-overlap threshold.

  Why this is correct

  Copyright regurgitation is a measurable overlap problem, so comparing generations against an indexed corpus of known works and blocking high-similarity spans directly targets the first risk. This is the mechanism behind output-side copyright filters and it produces a concrete, auditable threshold rather than a vague policy statement. Because the check runs on the generated text, it catches memorized passages regardless of how the prompt elicited them.

• ✓

  Add a named-entity and defamation classifier in the output rail that flags assertions about private individuals and routes them to human review before publication.

  Why this is correct

  Defamation risk concentrates on factual assertions about identifiable people, so an output-side classifier that detects named private individuals combined with claim-bearing language and escalates to a human reviewer addresses the second risk precisely. Human review is the accepted mitigation because automated truth assessment is unreliable. This control is complementary to the similarity filter rather than redundant with it.

• ✗

  Restrict the assistant to a smaller parameter model fine-tuned only on the company's own published archive.

  Why it's wrong here

  A smaller model fine-tuned on owned content reduces some risk but does not eliminate memorization, and the scenario's assistant is an existing NIM-hosted deployment. Swapping the base model is a substantial re-architecture that the question does not call for. It also leaves defamatory statements about private individuals entirely unaddressed, since those can be generated from any training corpus.