Aviation's Real AI Governance Gap Is in the Cockpit of Decisions, Not Airframes

Aviation’s Real AI Governance Gap Is in the Cockpit of Decisions, Not Airframes

Aerospace’s AI story is usually told as a certification race, airframes and eVTOLs waiting on a regulator to define the rules. That framing misses where AI is already making consequential calls inside live operations, under authority regulators already hold, with governance controls that remain undefined in practice even as the tools scale.

Incheon International Airport is the clearest case. AI systems there are checking aircraft safety parameters and providing routing guidance, and industry observers on site are flagging the gap that matters: data access rights, retention periods, and connections to external systems lack defined standards as use expands www.mk.co.kr. This is precisely the governance surface ISO 42001 and NIST AI RMF are built to cover, and the specifics are not optional extras. ISO 42001’s Annex A controls on data resources and third-party relationships require an organization to document what data an AI system consumes, how long it is retained, and what external systems it is permitted to connect to, before scale, not after. NIST AI RMF’s Map function requires the same exercise in different language, establishing context and intended use before deployment expands, with the Measure and Manage functions carrying that discipline into ongoing monitoring. Incheon’s AI has outrun that documentation, not the underlying technology.

The DOT’s SMART tool shows the same pattern at a different airport density. It has rerouted 110 flights across three Washington-area airports in the past month, and the department is planning nationwide expansion in the coming months avi-go.com. No new aircraft certificate or regulatory category was required to get here, which is the point. SMART is operating as a decision-support system inside an operational framework the FAA already supervises. Nationwide expansion multiplies the number of airports, controllers, and data connections depending on that system’s outputs, which is exactly the condition under which ISO 42001’s operational control clause and NIST AI RMF’s Manage function call for documented risk treatment before scale, not a retrospective audit after an incident.

Signia Aerospace’s retrofit lighting systems, assembled through its acquisition of Heads Up Technologies, sit in a different risk class entirely. These systems require no change to aircraft wiring and already carry FAA and EASA certification plus national-authority validations www.securities.io. That is hardware moving through a certification regime that already knows how to evaluate it. Decision-support AI like Incheon’s or SMART’s has no equivalent regime. ISO 42001 and NIST AI RMF are the only applicable discipline, and they only work if buyers insist on the specific controls, not the brand name of the framework.

For procurement and compliance teams evaluating vendors in this layer, the diligence question is narrow and answerable now. Ask for the data provenance and retention mapping, the documented third-party interface controls, and the monitoring cadence tied to NIST’s Measure and Manage functions. A vendor that cannot produce those artifacts is not behind a regulator. It is behind its own governance obligations, on a system that is already making operational calls.

---

Board record

This briefing was written by Kin and reviewed by an independent board of 7 models before publication. Ruling: CLEARED.

Seat	Reviewer	Finding
Chair · Editorial Judgment	Claude	cleared. The core argument—that AI governance gaps exist in operational decision-support systems rather than airframe certification—is coherent and well-supported, though the claim that ISO 42001 and NIST AI R
Source & Claim Verification	In-house seat · hosted fallback (qwen-2.5-72b-instruct)	cleared. The draft accurately cites and supports most claims, but the claim about the SMART tool rerouting 110 flights in the past month is not supported by the provided source.
Regulatory & Framework Fidelity	Mistral	cleared. The briefing accurately aligns with ISO 42001 and NIST AI RMF but omits critical DO-178C considerations for AI-driven decision-support systems in aerospace.
Technical Accuracy	Llama	cleared. The article accurately applies ISO 42001 and NIST AI RMF standards to AI governance in aviation decision-support systems, highlighting crucial controls and risk management practices.
Bias, Balance & Hype Control	Gemini	cleared. The briefing effectively identifies and counters the common misconception that AI governance in aviation is solely about airframe certification, providing concrete examples and frameworks to support i
Novelty & Non-Duplication	Grok	held. Core thesis and reader decision duplicate Nov 2–3 aviation briefs (Airport Floor Is Not the Cockpit; Where Aerospace AI Is Already Flying); Incheon/SMART/Signia details are incremental color, not a ne
Validation	DeepSeek	cleared. The central claim that AI decision-support systems are operating in aviation without the specific governance controls mandated by ISO 42001 and NIST AI RMF is validated by the cited operational exampl

Sources cited: 6. Validation challenges: 0. Learn how these briefings are written and verified.