1. n. [Digital]

An adversarial attack is a technique where intentionally crafted inputs are fed to AI models to deceive or mislead them into making incorrect predictions or classifications.

Relevance to business: Adversarial attacks highlights vulnerabilities in AI systems, necessitating robust security and testing to protect critical applications from exploitation, reduce risk, and ensure reliability.