← AI security in plain English

AML.T0074

MITRE ATLAS / Masquerading

Adversaries manipulate features of their artifacts so those artifacts appear legitimate or benign to users and security tools, including names, metadata, file properties, and repository presentation.

Think of it likePutting the real brand's label on a counterfeit tin so it passes on the shelf.

In plain English

A malicious model or package that carries a familiar name, a plausible author, and tidy documentation gets installed. Most reviews check the label rather than the contents.