An AI assurance glossary for aerospace and defense

These terms appear throughout aerospace and defense AI work. Each definition below is what the cited source says, not a restatement of industry slang. The same terms are collected as individual anchors, with DefinedTerm markup, on the /glossary/ index.

Status as of 14 October 2026: the FAA page is still Roadmap for Artificial Intelligence Safety Assurance, Version I. EASA Issue 2 remains the published guidance, dated 6 March 2024. CMMC Phase II remains suspended (13 July 2026) with Phase I self-assessment in place; see CMMC Level 2 vs Level 3 for AI systems.

Safety assurance and learning assurance

AI safety assurance. The FAA roadmap’s primary purpose is a path to assuring the safety of AI in aircraft and related systems for inflight operations. The same document says AI must demonstrate its safety before use in aviation.

Learned versus learning AI. The FAA distinguishes static (learned) from dynamic (learning) implementations. Learned AI is frozen after training; learning AI continues to change in operation and is treated as an ongoing research topic.

Learning assurance. EASA proposes a new concept of learning assurance to extend the traditional development assurance, aimed at confidence that trained models generalize on unseen operational data and are robust in foreseeable conditions.

W-shaped process. EASA illustrates learning assurance as a W-shaped cycle that adapts the typical development assurance V-cycle to ML concepts, running with the traditional V-cycle for non-ML constituents.

Operational design domain (ODD). EASA ties data quality to the ODD: a correct and complete definition of the ODD is a prerequisite to an adequate level of quality of the data sets.

Human-AI teaming (HAT). EASA’s Issue 2 introduces the concept of ‘human-AI teaming’ (HAT) for Level 2 systems that automatically take decisions under human oversight. See EASA AI Concept Paper levels explained.

Airborne software and data

DO-178C. RTCA describes DO-178() as the software standards documents used with EUROCAE ED-12 for airborne software development assurance. The FAA roadmap says it is supporting the development of voluntary consensus standards, including application and possible extension of DO-178C, DO-330 and related documents. See DO-178C and machine learning gaps.

DO-330. The same FAA roadmap names DO-330 with DO-178C as a document the FAA is supporting for application and possible extension. It is the tool-qualification consideration used with airborne software, not a model certificate.

Development assurance. EASA treats learning assurance as an extension of traditional development assurance, not a replacement of it.

Risk, governance and ethics frameworks

NIST AI RMF functions. The Core is composed of four functions: Govern, Map, Measure and Manage. NIST says the actions do not constitute a checklist.

TEVV. NIST AI RMF 1.0 discusses test, evaluation, verification, and validation (TEVV) as part of measuring AI risk. The framework does not turn those words into an aviation qualification method.

AI management system. ISO/IEC 42001 specifies requirements for an Artificial Intelligence Management System. ISO describes it as managing AI-related risks across an organization rather than looking at details of specific AI applications.

DoD AI ethical principles. The Department adopted five ethical principles for the use of AI in February 2020: Responsible, Equitable, Traceable, Reliable and Governable. See DoD RAI Toolkit vs NIST AI RMF.

High-risk AI system. In the EU AI Act, high-risk systems are those that meet the conditions in Article 6 and the Annex III use-case list, subject to the military, national-security and other scope rules in Article 2. See EU AI Act scope for defense and aviation.

Controlled data

Controlled Unclassified Information (CUI). NIST SP 800-171 Rev. 3 is the set of security requirements for protecting the confidentiality of CUI when that information is processed, stored or transmitted by nonfederal organizations.

CMMC. The CMMC Program in 32 CFR Part 170 is the Department of Defense assessment mechanism for contractor implementation of those CUI protections. It is not an AI product-assurance standard. See CMMC and AI with CUI and CMMC Level 2 vs Level 3 for AI systems.

DoD principles, one by one

Each word below is the Department’s February 2020 definition, not a NIST function.

Responsible. Department personnel will exercise appropriate levels of judgment and care while remaining responsible for the development, deployment, and use of AI capabilities.

Equitable. The Department will take deliberate steps to minimize unintended bias.

Traceable. Capabilities are developed and deployed so that personnel understand the technology, including transparent and auditable methodologies, data sources, and design procedures and documentation.

Reliable. Capabilities have explicit, well-defined uses, with safety, security and effectiveness subject to testing and assurance within those uses across their entire life-cycles.

Governable. Capabilities have the ability to detect and avoid unintended consequences, and to disengage or deactivate deployed systems that demonstrate unintended behavior.

RAI Strategy and Implementation Pathway. CDAO’s toolkit release says the June 2022 pathway spells out 64 lines of effort for operationalizing the Department of Defense’s AI Ethical Principles, including developing technical tools and guidance.

AIA Toolkit. The live Tradewinds page titles the public CDAO self-evaluation resource the AI Assurance (AIA) Toolkit. It is a self-evaluation resource that helps teams assess their AI projects at any stage of development.

Model card (CDAO generative-AI companion). The December 2024 CDAO Generative AI Responsible AI Toolkit lists a Hugging Face model-card template as a documentation tool for traceability (transparent and auditable documentation).

NIST functions and later profiles

Govern, Map, Measure, Manage. The Core is composed of four functions: Govern, Map, Measure and Manage. They are the same four names as the “NIST AI RMF functions” term above, listed individually so a search for one function still hits this glossary.

SSDF Community Profile (SP 800-218A). NIST augments the secure software development practices and tasks defined in SSDF version 1.1 with practices specific to AI model development. The additions are documented in the form of an SSDF Community Profile.

Dual-use foundation model. SP 800-218A is a companion resource to the SSDF to incorporate secure development practices for generative AI and for dual-use foundation models.

Adversarial machine learning (AML). NIST AI 100-2e2025 provides a taxonomy of concepts and defines terminology in the field of adversarial machine learning (AML).

Aviation development-assurance levels and data

FDAL and IDAL. FAA AC 20-174 says applicants should propose and document in your certification plan the functional development assurance level (FDAL), item development assurance level (IDAL) and justification for the levels.

IDAL A/B ML ceiling (EASA Issue 2). EASA will initially accept only applications where AI/ML constituents do not include IDAL A or B items for supervised learning.

Level 1 and Level 2 (EASA). Issue 2 is guidance for Level 1 & 2 machine learning applications. Level 2 is the human-AI teaming case on that news item.

Aeronautical data. AC 20-153B defines aeronautical data as data used for aeronautical applications such as navigation, flight planning, flight simulators, terrain awareness, and other purposes.

Letter of Acceptance (LOA). An LOA confirms compliance with RTCA/DO-200B.

Runtime assurance (RTA). NASA/TM–20220015734 defines RTA as the combination of RTM and one or more functions triggered by RTM, such as recovery, failover, warning alarm, or shutdown.

Annex III. In the EU AI Act, high-risk systems include those that meet Article 6 and the Annex III use-case list.

Article 2 scope. That high-risk list is subject to the military, national-security and other scope rules in Article 2.

How to use this list

Analysis, not sourced fact: if two sources use the same English word, check which document you are answering. “Assurance” in the FAA roadmap is safety assurance of an airborne function. “Assurance” in an ISO 42001 audit is evidence that a management system is operating. “Responsible” in the DoD principles is a named ethical principle, not a NIST function.