Knowledge Bases and Evaluation Organizations: MITRE ATLAS

中文版

Knowledge Bases and Evaluation Organizations · MITRE ATLAS. Knowledge base of adversarial threats to AI. This source has 4 entries admitted under this category's review policy; each links directly to the original.

Authority tiers in this category: A2–B. Admission policy: All core resources are included. Each secondary source is admitted according to its own database, methodology, versions and leaderboard structure; there is no cross-organization scoring scheme.

• MITRE ATLAS: knowledge base of adversarial threats to AI

  Publisher: MITRE · Collection: MITRE ATLAS · Format: Threat knowledge base and matrix · Authority tier: A2 · Review status: living

  Organizes adversarial behavior against AI systems into tactics, techniques, mitigations and case studies, covering predictive, generative and agentic AI.

  Evidence use: Supports classification of AI attack techniques, threat modeling and case mapping.

  Limitations: Continuously updated; citations should include the access date. Individual cases still require tracing the underlying evidence.

  Related concepts: AI Red Teaming · AI Governance and Regulation

  Link last checked: 2026-09-24

• MITRE ATLAS Fact Sheet

  Publisher: MITRE · Collection: MITRE ATLAS · Format: Official framework description · Authority tier: A2 · Review status: living

  Explains ATLAS's purpose, evidence base, relationship to ATT&CK and main use cases.

  Evidence use: Supports summarizing ATLAS's positioning, boundaries and uses.

  Limitations: The overview file does not replace the online matrix. Statistics follow the online version.

  Related concepts: AI Red Teaming

  Link last checked: 2026-09-24

• MITRE ATLAS Data

  Publisher: MITRE · Collection: MITRE ATLAS · Format: Versioned threat knowledge dataset · Authority tier: A2 · Review status: living

  Publishes ATLAS tactics, techniques, mitigations, relationships and case studies as versioned YAML, STIX and tabular data.

  Evidence use: Supports machine-readable analysis, version checking and complete citation of ATLAS's knowledge structure.

  Limitations: Continuously updated; citations should include the access date. Cases and mitigations must be validated against the specific system.

  Related concepts: AI Red Teaming · AI Guardrails

  Link last checked: 2026-09-24

• MITRE SAFE-AI Framework

  Publisher: MITRE · Collection: MITRE ATLAS · Format: AI security engineering framework · Authority tier: A2 · Review status: living

  Turns ATLAS threat knowledge into security engineering activities across the AI system lifecycle.

  Evidence use: Supports bringing adversarial threats into design, testing, deployment and operations.

  Limitations: Continuously updated; citations should include the access date. Cases and mitigations must be validated against the specific system.

  Related concepts: AI Red Teaming · AI Governance and Regulation

  Link last checked: 2026-09-24