AML.T0102

MITRE ATLAS / Generate Malicious Commands

Adversaries use language models to generate malicious commands dynamically from natural language, producing payloads at runtime rather than shipping fixed code.

Think of it likeBringing along someone who can improvise a new tool on the spot instead of packing every tool in advance.

In plain English

Malware that writes its own commands as it goes has no fixed strings to match. Each run produces something slightly different, which defeats signature-based detection.