The Autonomy Gap Is Not the Risk You Should Be Managing

EASA has said, in public, that it is behind on AI, and it frames the gap around autonomy. Self-driving cars are on the road, autonomous aircraft are not close, and the agency does not expect to close that distance soon aviationweek.com. Its certification ladder treats full autonomy as the hard problem still years out www.instagram.com. The FAA’s contribution at EASA AI Days makes the same point from the assurance side: aviation has little track record certifying AI safety, and the technology is outrunning the evidence base regulators would need to sign off on it www.easa.europa.eu.

That diagnosis is accurate. It is also describing a different problem than the one sitting in your supply chain this quarter. Full autonomy is a rulemaking horizon. AI embedded in maintenance workflows, requirements traceability, and design verification is a live deployment already underway. Defense programs generate hundreds of thousands of interconnected life-cycle requirements that cannot be reviewed manually, and AI tools are being brought in specifically to manage that load, explicitly as augmentation to engineering judgment rather than a replacement for it militaryembedded.com. The DER and ODA structures named in postings like Gulfstream’s Airworthiness Engineer role still hold the pen on certification documentation careers.gulfstream.com. Nobody is proposing to automate that signature away. But the drafts, matrices, and traceability packages landing on those engineers’ desks are increasingly AI-assisted, and no EASA or FAA standard currently states how much confidence that assistance needs to carry before it enters the record.

What the MAA model actually offers

The UK’s Military Aviation Authority does not treat AI as a new category of system. It treats AI output the way it treats any engineering artifact: assured to a defined confidence level regardless of how it was produced institute.global. That is not a new regulatory invention. It is the same logic DER and ODA reviewers already apply to every other artifact entering a certification file, where review burden tracks the criticality of the claim, not the origin of the document. The useful move here is not that the MAA discovered something commercial aviation lacks. It is that the MAA has stated plainly, for AI specifically, a principle EASA’s rulemaking track has not yet applied to it. EASA’s public materials are still organized around classifying systems by degree of autonomy, a category question that says nothing about whether a specific AI-generated traceability matrix deserves scrutiny today www.easa.europa.eu.

These are two separate gaps. One is EASA and the FAA lacking a certification path for autonomous flight. The other is that AI-assisted engineering artifacts are already moving through existing DER and ODA review without anyone having named the confidence level those artifacts should meet. Closing the first gap is a multi-year rulemaking effort. Closing the second is a supplier-management question a program office can act on now.

The near-term decision is not whether to wait on EASA or the FAA to rule on autonomous flight. It is whether your suppliers’ AI-assisted tools are being held to any stated confidence standard, or whether that question has simply not been asked yet. Human oversight of AI in defense and aerospace systems has drawn attention as high as the UK monarchy www.adsadvance.co.uk. It should draw yours before the next audit does.

---

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. Core argument distinguishing autonomy-gap from artifact-confidence-gap is coherent and well-supported, but the claim that ‘no EASA or FAA standard currently states how much confidence AI assistance ne
Source & Claim Verification	Qwen · local	cleared. Several claims are not directly supported by the provided sources, particularly those regarding the FAA’s and EASA’s positions on AI certification and the specific practices of DER and ODA reviewers.
Regulatory & Framework Fidelity	Mistral	held. seat error: Client error ‘404 Not Found’ for url ‘https://openrouter.ai/api/v1/chat/completions’
For more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/404
Technical Accuracy	Llama	cleared. The article accurately distinguishes between the regulatory gap for autonomous flight and the immediate need for confidence standards in AI-assisted engineering artifacts.
Bias, Balance & Hype Control	Gemini	cleared. The briefing effectively distinguishes between two distinct problems, directly addressing and refuting a potential counterargument about the relevance of autonomy in the near term, and avoids vendor h
Novelty & Non-Duplication	Grok	cleared. The core reframing—autonomy rulemaking as a multi-year distraction versus the live, ungoverned flow of AI-assisted DER/ODA artifacts as a supplier-management gap actionable now—is a sharp synthesis no
Validation	DeepSeek	cleared. The central claim that AI-assisted engineering artifacts are moving through certification review without a stated confidence standard is an unverified assertion, not a fact that can be adversarially t

Sources cited: 13. Validation challenges: 0. Review cost: about $0.04. Learn how these briefings are written and verified.