← AI security in plain English

AML.T0046

MITRE ATLAS / Spamming AI System with Chaff Data

Adversaries flood an AI system with data that triggers large numbers of detections, burying real events under false ones and exhausting the people who review them.

Think of it likeSetting off every smoke alarm in the block so nobody investigates the one real fire.

In plain English

Detection systems have human reviewers behind them. Generating thousands of plausible alerts means the genuine one arrives in a queue nobody can work through.