ATLASAML.T0031
ATLAS index
AML.T0031

Erode AI Model Integrity

Adversaries may degrade the target model's performance with adversarial data inputs to erode confidence in the system over time. This can lead to the victim organization wasting time and money both attempting to fix the system and performing the tasks it was meant to automate by hand.

Framework
MITRE ATLAS
Maturity
Realized
Platforms
Predictive AI, Generative AI, Agentic AI
Release
2026.05

Overview

Adversaries may degrade the target model's performance with adversarial data inputs to erode confidence in the system over time. This can lead to the victim organization wasting time and money both attempting to fix the system and performing the tasks it was meant to automate by hand.

Sources

  1. MITRE ATLAS AML.T0031: Erode AI Model Integrity — MITRE