MITRE ATLAS framework

concept brazil Updated Sep 10, 2026 1 min read about-mitre-atlas.md
Raw file servicenow.com

The MITRE ATLAS framework documents adversarial tactics and techniques targeting AI (artificial intelligence) and ML (machine learning) systems, complementing traditional IT security frameworks.

Framework overview

The MITRE ATLAS (Adversarial Threat Landscape for Artificial-Intelligence Systems) framework provides a structured knowledge base for understanding and defending against attacks on AI and ML systems.

It complements the MITRE-ATT&CK framework, which documents threats to traditional IT infrastructure.

Note: AI systems may produce inaccurate results. Implement human review processes when working with AI-generated data.

Integration with Security Operations

MITRE ATLAS techniques and tactics are ingested, viewed, and associated with security incidents using the same MITRE-ATT&CK Repository, MITRE-ATT&CK Card, and association workflow as MITRE-ATT&CK.

MITRE ATLAS is distinguished from MITRE-ATT&CK by an AML-prefixed technique or tactic ID.

For example, a technique ID might be AML.T0000, and a tactic ID might be AML.TA0002.

Technique names can collide between the two frameworks, so the ID is the reliable indicator. There is no separate object type for MITRE ATLAS.

For setup steps, see Activate the MITRE ATLAS framework.

TAXII collection

Unlike MITRE-ATT&CK, which exposes three TAXII collections (Enterprise, Mobile, and ICS), MITRE ATLAS exposes only one TAXII collection ("ATLAS").

STIX object mapping

MITRE ATLAS techniques and tactics are ingested using the same STIX object types as MITRE-ATT&CK (for example, Attack Pattern for techniques and Course of Action for mitigations).

The Data Source and Detection fields available for MITRE-ATT&CK techniques are not populated for MITRE ATLAS techniques.

Keywords MITRE ATLASAI securitymachine learning threats
Search documentation
Search document titles and file paths on the brazil branch