The FAA just handed Boeing back a power it took away after two fatal crashes. As reported via Hacker News on Friday, the U.S. government cleared Boeing to once again issue airworthiness certificates for its 737 MAX and 787 Dreamliner aircraft. That’s the final sign-off saying a plane is safe to fly before it goes to the customer.
Boeing lost that authority after the 2018 Lion Air and 2019 Ethiopian Airlines crashes. Now it’s back.
What Actually Changed
- Before September 2025: FAA inspectors personally signed off on every MAX and Dreamliner leaving the factory.
- September 2025 to now: A split arrangement. Boeing and the FAA alternated weeks doing the certification work on a portion of the fleet.
- As of Friday: Boeing does all of it again, with FAA oversight sitting above the process rather than inside it.
The FAA’s reasoning was blunt and data-driven. “During the past eight months, the FAA has seen comparable production quality findings when Boeing issued airworthiness certificates and when the FAA issued them,” the agency said. “Based on these results, the FAA determined it can safely return this responsibility to Boeing.”
Boeing’s response was measured. The company said it “will continue to work under the oversight of the FAA in building safe, high-quality commercial airplanes that comply with all airworthiness certification requirements.”
Why This Lands on an AI Reader’s Desk
This isn’t an AI story. But it’s the clearest live case study of a question the AI industry is arguing about right now: when do you let a company certify its own product?
That’s the whole debate around AI model evaluations. Labs run their own safety tests, publish their own system cards, and grade their own capability thresholds. Critics call it self-certification with extra steps. Defenders point out that regulators lack the technical staff to do it themselves.
Aviation went through this exact arc, and the timeline is worth internalizing:
- Delegated authority is the default. The FAA never had the headcount to inspect every aircraft. Delegation was normal, not exceptional.
- Failure triggers clawback. 346 deaths ended the arrangement. Not a policy debate. A body count.
- Restoration requires measured parity. Boeing didn’t get this back on promises. It got it back because the FAA ran a shadow comparison for eight months and found no meaningful gap between Boeing’s findings and its own.
What stands out here is step three. The FAA built an A/B test into the regulatory process. Alternating weeks between regulator and company created a control group. That’s a mechanism the AI world mostly doesn’t have.
The Trust Timeline Is Long
Seven years from the first crash to full restoration of certification authority. That includes a 20-month grounding, a criminal fraud settlement, congressional hearings, and a door plug blowing off a 737 MAX 9 mid-flight in January 2024, which reset the clock again just as things were stabilizing.
For anyone building AI products in regulated or safety-adjacent spaces, that’s the realistic timeline for rebuilding institutional trust after a serious failure. Not a quarter. Not a year. Years, plus a demonstrated data trail.
What to Expect Next
Boeing is one of the largest U.S. exporters by value, and delivery speed is the company’s main financial lever. Expect production rates to climb now that the certification bottleneck is gone.
Watch for two things:
- Whether quality findings hold. The FAA’s justification was statistical parity. If Boeing’s numbers drift once it’s grading solo, that becomes the story.
- Whether other regulators copy the mechanism. The alternating-weeks comparison is a genuinely good design. Anyone drafting AI oversight rules should be reading it.
The useful lesson isn’t that self-certification works or doesn’t. It’s that the FAA found a way to measure whether it was working before committing. That beats trusting a press release, and it beats a blanket ban.
More detail on the FAA’s decision and Boeing’s production outlook is available at the original source.