A risk manager is assessing risks specific to retrieval-augmented generation (RAG) systems. Which risk is MOST unique to RAG architectures compared to standard LLM deployments?
Select an answer to reveal the explanation.
Short Explanation
You're right to pick the retrieval-corpus access-control risk because RAG pulls from external documents, and weak permissions can leak privileged content to unauthorized users. The other risks are generic LLM quality, refusal, or cost issues.
Full Explanation
The correct risk is that sensitive documents in the retrieval corpus may be surfaced to unauthorized users because RAG retrieves from an external corpus, so weak retrieval access controls can leak privileged content. Grammatical errors are a general LLM quality risk. Refusing questions outside the training domain is a non-RAG limitation that RAG addresses. High inference costs from large parameter counts are a general LLM infrastructure concern.