Ethical Robustness

Model Evasion

Attack strategy where specially crafted inputs allow bypassing the detection or classification mechanisms of an AI model. Evasion directly threatens ethical robustness by allowing the violation of established rules and moral constraints.

← Indietro