Model Robustness

Extraction attacks

Attack strategy aiming to reproduce or steal a proprietary model by querying its API and analyzing its responses. These attacks exploit information leaks through predictions to reconstruct the model or its training data.

← Wstecz