ScutOS

AI & LLM Threat Intelligence

A dedicated mapping of AI-centric threats combining the MITRE ATLAS framework and the official OWASP Top 10 for LLM Applications (2025).

Adversarial Threat Landscape for AI
Visualizing the adversary-centric lifecycle against AI/ML systems
AML.TA0002

Reconnaissance

Gathering information about the AI system

AML.TA0003

Resource Development

Setting up resources to target the AI system

AML.TA0001

Initial Access

Gaining access to the AI system or environment

AML.TA0004

ML Model Access

Gaining read or write access to the ML model

AML.TA0005

Execution

Running malicious code or inputs within the AI environment

AML.TA0006

Persistence

Maintaining access to the AI system across restarts

AML.TA0007

Privilege Escalation

Gaining higher-level permissions in the AI environment

AML.TA0008

Defense Evasion

Avoiding detection while attacking the AI system

AML.TA0009

Credential Access

Stealing credentials used by the AI system

AML.TA0010

Discovery

Learning the internal structure of the AI system

AML.TA0011

Collection

Gathering data, model weights, or artifacts

AML.TA0012

ML Attack Staging

Staging adversarial inputs or poisoned data

AML.TA0013

Exfiltration

Stealing the model, datasets, or intel

AML.TA0014

Impact

Degrading, manipulating, or denying the AI service