MITRE ATLAS · AI Attack Adaptation
AML.T0043 · Craft Adversarial Data
Adversarial data are inputs to an AI model that have been modified such that they cause the adversary's desired effect in the target model. Effects can range from misclassification, to missed detections, to maximizing energy consumption. Typically, the modification is constrained in magnitude or location so that a human still perceives the data as if it were unmodified, but human perceptibility ma
Mitigations
- Predictive AI Output Obfuscation
- Predictive AI Model Hardening
- Limit AI Service Query Volume and Rate
- Predictive AI Ensembles
- Predictive AI Input Restoration
- Predictive AI Adversarial Input Detection
In the news
-
A Wolf in Sheep's Pixels: Visual Camouflage Attacks via HDR Gain Maps
DRJanus is a visual camouflage attack that embeds a harmful image in a JPEG's HDR gain map, evading automated checks that read only the SDR base image.
ReviewReview image moderation pipelines that rely only on SDR image content
AML.T0043 arXiv:2611.25469 #content-moderation#jpeg#adversarial-images 1 min read Brief In your stack
Tools
Test or exercise it
- Adversarial Robustness Toolbox LLM red-teaming · Trial · Dual-use
- Counterfit LLM red-teaming · Hold · Dual-use
Detect or mitigate it
- Adversarial Robustness Toolbox LLM red-teaming · Trial
- Counterfit LLM red-teaming · Hold
Mappings are AI-assisted and checked against the framework definitions; corrections welcome.