0 alternative definitions
An AI system's capacity to maintain reliable performance when facing intentionally crafted inputs designed to mislead or degrade its outputs.
