← AI security in plain English

AML.TA0013

MITRE ATLAS / Credential Access

The adversary tries to steal account names and passwords, using techniques such as Unsecured Credentials, RAG Credential Harvesting, and Credentials from AI Agent Configuration.

Think of it likeRifling through the kitchen drawers for the sticky note where someone wrote down all the passwords.

In plain English

Attackers hunt for logins and keys, pulling secrets out of AI agent configs, knowledge bases, or carelessly stored credentials to unlock more of the system.