A large language model deployed as a customer service chatbot occasionally generates factually incorrect responses presented with high confidence. Which emerging AI risk category does this BEST represent?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Here's the deal — b is correct because hallucination refers to the tendency of large language models to generate confident, fluent, but factually incorrect or fabricated content — a distinct risk category for LLM deployments in customer-facing applications. A is an attack vector.
Full explanation below image
Full Explanation
B is correct because hallucination refers to the tendency of large language models to generate confident, fluent, but factually incorrect or fabricated content — a distinct risk category for LLM deployments in customer-facing applications. A is an attack vector. C is a model theft attack. D is a privacy risk from memorization.