About...

Courts have long known the difference between what matters and what is said in passing.

The ratio decidendi — the binding core of a judgment - is what lawyers argue about, what textbooks cite and what politicians invoke when they need authority without accountability.

The obiter dictum is everything else. Side observations. Passing remarks. Technically irrelevant.

Also, frequently, the part that turns out to matter.

This is obliter. The extra letter is intentional. Obiter observations merely pass by; obliter ones have a tendency to obliterate the assumptions they land on. This is either a legal pun of questionable taste or an accurate description of what happens here. Possibly both.

Why this exists

Something interesting happened when artificial intelligence met European regulation.

The interesting thing was not the technology - technology is always interesting, in the way that a very large object rolling toward you is interesting. The interesting thing was the gap between what the regulation said, what it meant, and who it would actually benefit.

This gap is well-documented. It lives in recitals, in subordinate clauses, in definitions buried on page 47 of a 100-page framework document that everyone has cited and nobody has read.

obliter exists to read page 47.

The Clause

A word about a recurring presence.

Throughout these pages, you will encounter The Clause. It is, depending on context, a specific provision of applicable law, the collective noun for those who draft such provisions, or the reason the answer to your question is longer than you hoped.

The Clause did not create the problems examined here. It merely appears, reliably, wherever those problems are most comfortable. It has not endorsed this publication - it rarely endorses anything, since endorsements create obligations, and The Clause has learned to prefer arrangements that do not.

It is not a villain. Villainy requires intention, and The Clause operates at a level of abstraction where intention becomes a category error. It is simply present. It was present when the AI Act was drafted. It was present when the compliance cost estimates were written. It will be present when the appeals are filed.

obliter is, among other things, an attempt to document its movements.

Five questions we keep returning to...

Not because we have answers. Because the questions themselves are under-examined - and under-examination tends to be where The Clause does its best work.

The Regulator's Gift
The EU AI Act imposes rigorous compliance requirements on high-risk AI systems. For a company with forty lawyers and a dedicated compliance team, this is an afternoon's work. For a four-person startup trying to build a medical diagnostic tool, it is a reason to stop. The question is whether a regulation designed to constrain Big Tech has instead handed it the market.

The Invisible Handshake
Classic competition law requires a meeting of minds - people in a room, agreeing to do something they shouldn't. What happens when the agreement is never made, because the parties involved are algorithms optimising the same objective function and independently concluding that competition is inefficient? No conspiracy. No communication. Just a Nash equilibrium that looks, from the outside, remarkably like a cartel.

Cognitive Real Estate
Your attention is not free. Your decision-making is not neutral. Behavioural economics has known this for decades; what it did not anticipate was an entity that knows not just the general shape of human cognitive bias, but your specific bias, your current emotional state, and the precise heuristic most likely to produce the outcome it prefers. Article 5 of the EU AI Act prohibits manipulative AI practices. The question is what manipulation means when the manipulation is indistinguishable from a very good conversation.

The Monoculture Risk
Nassim Taleb's Black Swan arrives from outside the distribution of observed events. AI systems, being statistical machines, are constitutionally unable to prepare for it. If the financial sector, the logistics sector and the healthcare sector all rely on variants of the same two or three foundational models, they share not just infrastructure but cognitive blind spots. When the Swan arrives, they will all be surprised in exactly the same way, at exactly the same moment.

The Accountability Sink

An accountability sink is a structure that absorbs responsibility the way a heat sink absorbs warmth - taking it in, spreading it across so many surfaces that no single point is ever hot enough to touch. Bureaucracies have built them for centuries. What is new is that we can now manufacture one out of software and deploy it at the precise moment a decision touches someone's liberty, money, or rights. The question this research area tracks is the one every other question on this site quietly assumes an answer to: when an AI system acts and harm follows, who is responsible? Not "which company will eventually be sued" - that question has answers, slow and unsatisfying ones. 

...but small questions are ones we'll keep looking for

obliter is first and foremost a repository. Not of answers - the questions above are not the kind that resolve cleanly - but of the thinking that accumulates around them: arguments tested, frameworks that hold and those that don't, the moments when something in the news obliterates a comfortable assumption and the work of rebuilding begins.

The large questions are listed above. They are not, however, the primary object of interest here. The large questions are well-attended - there is no shortage of people explaining, at length, what AI means for democracy, for labour, for the future of everything. The small questions are less popular. They require specificity. They resist the kind of thinking that scales well on LinkedIn.

Small questions are also where the actual work happens. Large questions carry too much weight to move freely - they are load-bearing for careers, for policy documents, for the reputations of people who have already staked a position. Small questions are under no such obligation. They can be followed wherever they lead. The useful insight tends to arrive sideways. Obliquely. Often in a footnote.

The Clause, it turns out, generates an extraordinary number of small questions simply by moving. Every new provision, every compliance deadline, every definition that quietly shifts between draft and final text produces a cluster of technical, overlooked, apparently minor questions that nobody is paid to ask. obliter is, among other things, an attempt to document those movements - and to ask the questions they leave behind.

Who writes this

I am the kind of person who finds small questions more interesting than large ones - not out of modesty, but from the conviction that specific, inconvenient questions do considerably more damage to comfortable arrangements than grand ones. Grand questions are easy to absorb. Small ones have a tendency to get stuck.

I work at the intersection of EU regulatory law, AI governance, competition policy and technology - an intersection that has no clean name yet, which is either a problem or an opportunity depending on your appetite for ambiguity. Whether the label is lawyer or technologist depends largely on who is asking and what they want to hear. The distinction, at this particular crossroads, is increasingly academic.

obliter is not affiliated with any institution. It publishes when something is worth saying and stops when it isn't.