§43 · Lane 8 — Agent Infrastructure Standards & Toolchain
MITRE ATLAS
The MITRE Corporation (2026) · MITRE ATLAS
Bibliographic data
- Title
- MITRE ATLAS — Adversarial Threat Landscape for Artificial-Intelligence Systems
- Authors / Issuing body
- The MITRE Corporation
- Venue / Publisher
- The MITRE Corporation
- Year
- 2026
- Designation
- standards-framework
- Licence
- Apache-2.0 — refer to publisher for full licence terms.
- Canonical link
- https://atlas.mitre.org/
How to cite
The MITRE Corporation (2026). MITRE ATLAS — Adversarial Threat Landscape for Artificial-Intelligence Systems. The MITRE Corporation. https://atlas.mitre.org/.
MITRE knowledge base of real-world adversary tactics and techniques against AI-enabled systems, modelled on the ATT&CK structure. Current data: 1 matrix, 16 tactics, 84 techniques, 56 sub-techniques, 32 mitigations, 42 case studies.
Why it matters for NETEVO
MITRE ATLAS is the threat-modelling layer the Agent Infrastructure Whitepaper standards-mapping table cites. Where OWASP names failure modes, ATLAS names adversary tradecraft — what an attacker does to realise a failure mode against an AI-enabled system. Citing ATLAS gives NETEVO surfaces a vocabulary for the adversarial side of the threat narrative that complements OWASP's defensive-failure-mode vocabulary.
Individual ATLAS technique identifiers (for example AML.T0018 Backdoor ML Model and the agentic-AI techniques added through early 2026) are cited in NETEVO prose by the catalogue's stable URL, not as separate entries. This single-entry convention mirrors MITRE CWE.
ATLAS pairs with the NIST AI RMF MEASURE function at the framework layer: NIST RMF tells the client what to measure and manage; ATLAS tells them what adversaries are likely to do. NETEVO tracks ATLAS monthly — the matrix is in an active expansion phase.
Where NETEVO applies this
- Agent Infrastructure Whitepaper — load-bearing — adversarial-tradecraft layer paired with OWASP failure modes