Products · Eur-Lex
Regulation × AI
Regulation in relation to AI is hard because you need grip on two layers at once — not a single snapshot in time.
Two layers of grip
When you deploy AI in a regulated domain, knowing “which rules apply now” is not enough. You need grip on:
- The applicable certification basis at the time approval was granted — which texts, versions, and interpretations formed the basis for that decision.
- The relevant updates on top — regulation is amended, supplemented with guidance, and corrected. What has changed since, and what applies to your situation?
Why systems and traceability matter
AI makes it easier to generate answers quickly. It does not make it easier to justify why that answer was valid at that moment — and which sources, versions, and decisions it rested on.
That is why Eur-Lex is not about “the AI knows all the law”. It is about traceability: which regulatory text applied when, which changes came on top, and how you hold that line in systems that must remain correct under oversight.
This is not legal advice — it is the systems engineering side of regulated AI: structure, versions, and an audit trail that holds up.
Jasper applied these same principles — certification basis, updates on top, and traceability where AI meets regulation — to the Aviation.bot concept, working it out with a software developer. The Aviation.bot concept still exists, but data ownership turned out to be a major barrier to further scaling.
How does data ownership play out in your organization in relation to AI?