The Airport Floor Is Not the Cockpit

Incheon International Airport is running an AI video analysis system that flags hazardous situations and abnormal signs during air traffic control, identifying aircraft and obstacles in real time MK. It is a sensible deployment. It is also a compliance category that most aerospace buyers are not yet tracking with the right framework.

The instinct in aviation is to route every AI-adjacent safety question through the avionics certification lens, DO-178C, FAA, EASA type approval. That lens is correct for software that flies on the aircraft. It is the wrong lens for software that watches the ramp, the runway, and the ATC console from the ground. Ground-based safety analytics like Incheon’s system are not airborne software. They do not go through a type certificate. They sit instead in airport and air navigation service provider operations, governed by safety management system obligations and, for any EU-connected deployment, by the EU AI Act’s high-risk classification for AI used as a safety component in transport infrastructure.

That classification matters because it triggers a specific, enforceable set of obligations, risk management systems, data governance, logging, human oversight, and conformity assessment, that exist independently of anything EASA or the FAA will say about airborne systems. And EASA’s bandwidth to say anything new, anywhere, is thin. The agency operates with roughly a quarter of the FAA’s certification staff, 280 full-time equivalents against the FAA’s 1,400, a gap that industry groups are now flagging explicitly as a scaling problem for European aeronautics ASD Europe. Aviation-specific guidance on AI in ATC and airport operations will not arrive quickly from that direction. The EU AI Act, by contrast, is already live law with its own timeline, and it does not wait for sector regulators to catch up.

For a compliance leader at an airport operator, an ANSP, or a vendor selling hazard-detection analytics into European airports, the decision point is concrete. Do not wait for an aviation-specific AI certification scheme to mature before building a governance file. Build the EU AI Act high-risk dossier now, risk management documentation, technical file, human oversight design, post-market monitoring, and layer ISO 42001 management-system controls underneath it to demonstrate organizational maturity to both regulators and customers. NIST AI RMF mapping is useful as a working methodology for the risk assessment itself, particularly for US-based vendors selling into airports that will eventually face parallel EU buyer due diligence.

The broader point is classification discipline. Aerospace AI governance risks collapsing into a single mental model, airborne software means DO-178C, full stop. Ground operations, ATC support tools, and airport safety analytics are a different system boundary with a different, already-binding legal regime. Getting that boundary wrong means either over-engineering a DO-178C-style assurance case where it is not required, or missing an EU AI Act high-risk obligation that carries real enforcement teeth. Buyers evaluating vendors in this space should ask which regime the product actually falls under before asking which certification the vendor claims to be pursuing.

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 ground-based airport AI falls under EU AI Act high-risk classification rather than DO-178C avionics certification—is logically sound and the regulatory distinction is real, thou
Source & Claim Verification	In-house seat · hosted fallback (qwen-2.5-72b-instruct)	cleared. The draft accurately cites and supports most claims, but the Instagram and YouTube sources are irrelevant and do not support any claims in the text.
Regulatory & Framework Fidelity	Mistral	cleared. The briefing accurately distinguishes between DO-178C (airborne systems) and EU AI Act high-risk obligations (ground-based ATC/airport operations), aligning with the correct regulatory frameworks.
Technical Accuracy	Llama	cleared. The article accurately distinguishes between the regulatory frameworks applicable to airborne software and ground-based safety analytics in airport operations, correctly identifying the EU AI Act as a
Bias, Balance & Hype Control	Gemini	cleared. The briefing effectively anticipates and counters the primary counterargument (DO-178C applicability) by clearly delineating system boundaries and highlighting the EU AI Act’s immediate relevance, whi
Novelty & Non-Duplication	Grok	held. Restates the 2026-11-02 briefing (Incheon airport AI + EASA staffing gap) with only a reframing of ground-vs-airborne classification, not a new fact, development, or decision.
Validation	DeepSeek	cleared. The DO-178C/airborne boundary is sound, but the EU AI Act high-risk claim is overbroad because airport/ATC video analytics are not automatically high-risk and require an Annex I/Annex III applicabilit

Sources cited: 11. Validation challenges: 0. Learn how these briefings are written and verified.