Languages: English | Français
The ability of an attacker to extract the training data of a generative model by prompting the model with specific inputs.
Source: NIST AI 100-2e2025 | Category: