Prepare and test your skills
Prepare and test your skills
Free sample question, answered and explained below. The Generative AI Leader practice bank has 2,191 more like it. Try 10 free: instant score, no account.
An enterprise customer service department is deploying a generative AI virtual assistant to handle complex customer inquiries. To improve response accuracy, the engineering team fine-tunes a foundational large language model using historical support tickets, email logs, and customer chat transcripts containing resolution details.
During security testing, engineers discover that specific prompt queries cause the model to output verbatim excerpts of unredacted customer names, home addresses, and private account details that were present in the historical support transcripts.
Which generative AI privacy vulnerability does this scenario demonstrate?
Model hallucination refers to the tendency of large language models to generate plausible-sounding but factually incorrect, fabricated, or nonsensical information that has no basis in the training data or grounded context.
Data drift (or distribution shift) refers to the phenomenon where the statistical properties and distribution of input data change over time compared to the data the model was originally trained on, leading to degraded model performance.
Prompt token truncation occurs when an input prompt or context payload exceeds the maximum context window limit supported by the model architecture, resulting in the excess tokens being cut off or discarded before inference.
Training data memorization (often referred to as data leakage or extraction) occurs when a machine learning model inadvertently encodes and stores exact verbatim sequences from its training or fine-tuning datasets within its internal model parameters. During inference, specific or adversarial prompts can cause the generative AI model to regurgitate sensitive, private, or proprietary data directly into its generated outputs.

Unlike standard algorithmic errors or hallucinations, training data memorization describes the precise phenomenon of a model learning and exposing real, underlying training records verbatim. Understanding this vulnerability is essential for establishing responsible AI governance, sanitizing training pipelines, and safeguarding confidential enterprise data.
Answer them yourself and get your score the moment you finish. No account needed.
Full access includes
Keep the momentum going with these hand-picked practice scenarios
Last updated