The Certification Basis That Won't Travel

Same Aircraft, Three Clocks

Bombardier’s Global 8000 received Transport Canada type certification in November 2025, U.S. FAA certification in December 2025, and EASA certification in January 2026, with certification in other jurisdictions still ongoing, according to Bombardier’s own disclosure. That is a conventional aircraft, with a stable certification basis, moving through three of the world’s most experienced regulators, and it still took months to clear each one in sequence.

For compliance leaders, that lag is a familiar cost of doing business. It becomes a harder problem once the certification basis itself stops being stable across jurisdictions, which is the direction AI-enabled avionics and autonomy systems are heading now.

One Regulator Has Already Moved

EASA is out ahead here. Its AI Roadmap 2.0 defines three general levels of AI involvement, from human assistance through human-AI teaming to advanced automation, and makes the degree of human oversight retained at each level the central question its guidance has to answer, as detailed in Aviation Business Middle East’s analysis. That is a structured, published, leveled framework. It gives programs something concrete to design toward, at least for the European market. Nothing comparable exists yet for the FAA or Transport Canada, and there is no guarantee the eventual equivalents will use the same three levels, the same oversight thresholds, or the same evidentiary bar.

The Convergence Counterargument

The optimistic case is that aerospace does not stay fragmented. Quality Magazine’s argument for integrated assurance ecosystems is built on exactly that premise: that the industry’s independent, siloed quality management approach is already under pressure to consolidate into shared assurance frameworks across suppliers and programs. If that logic extends to AI oversight, regulators converge on common levels the way they’ve converged on other airworthiness baselines over decades. The AIAA’s own framing of the challenge, that artificial intelligence spans a wide family of methods for perception, reasoning, and decision support, with machine learning as one subset fitting models to data rather than specifying rules directly, laid out in Aerospace America’s piece on scaling autonomy, carries a title, “build faster, learn together,” that is itself a bet on convergence. But that convergence is aspirational today, not operational. Nothing currently binds EASA’s levels to whatever the FAA eventually publishes.

The Gap That Matters

This is the decision point programs are not yet pricing in. A conventional, mechanically well-understood aircraft like the Global 8000 already needs a separate certification cycle in each jurisdiction. An AI-enabled system whose oversight requirements are defined level by level, and defined differently by each authority, cannot assume that sequence compresses. It extends it, because there is no shared basis to carry forward from one regulator to the next.

Cross-sector standards like ISO 42001 and the EU AI Act’s risk-tiering exist precisely to give industries a common documentation baseline before regulators fragment further. Aviation’s leveled frameworks are a preview of what happens when that baseline hasn’t arrived yet. Programs building AI-enabled systems now should assume today’s certification timeline is the floor, not the ceiling, and design their evidence packages to be legible to a framework that doesn’t exist yet, rather than optimized for the one regulator that has already published.

---

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 divergent AI certification frameworks will extend, not compress, multi-jurisdiction timelines—is logically sound and well-structured, but the leap from ‘EASA has published level
Source & Claim Verification	Qwen · local	cleared. All factual claims are supported by citations, but some sources are not directly relevant to the claims they are cited for, which could be improved.
Regulatory & Framework Fidelity	Mistral	cleared. The briefing correctly identifies ISO 42001 and the EU AI Act as cross-sector standards but does not substantively analyze or align its claims with their specific requirements or provisions.
Technical Accuracy	Llama	cleared. The article is generally technically accurate in its discussion of AI-enabled avionics and autonomy systems, but could be improved with more specific technical details on AI certification.
Bias, Balance & Hype Control	Gemini	cleared. The briefing effectively presents a counterargument and addresses potential vendor hype by highlighting the aspirational nature of convergence versus current operational realities.
Novelty & Non-Duplication	Grok	held. Timely Global 8000 hook plus EASA-AI framing yields a serviceable synthesis, yet the core claim that non-harmonized AI oversight will stretch multi-regulator timelines simply restates widely available
Validation	DeepSeek	cleared. The central claim that AI-enabled systems will face extended, non-convergent certification timelines is a plausible projection but cannot be validated against current reality, as the divergent regulat

Sources cited: 7. Validation challenges: 0. Review cost: about $0.04. Learn how these briefings are written and verified.