Europe debates AI regulation to manage global risks
The field notes for late August 2026 record a curious seasonal migration: Europe’s lawmakers, dispersed across their national capitals for the customary recess, reconvene in Brussels and Strasbourg to resume what has become an annual rite - the deliberation over how artificial intelligence ought to be governed. The observer arriving without prior knowledge of European institutional custom would note that this deliberation has been conducted, in one form or another, for several successive legislative seasons, and would reasonably ask what distinguishes this round of debate from the last, and whether the distinguishing feature is substantive or ceremonial.
The stated purpose of the exercise is productive in the fullest Veblenian sense: to manage risk and secure benefit from a technology whose consequences, we are told, will be felt globally, in every economy and society that has occasion to use a large model or be evaluated by one. This is not a trivial claim, and one does not doubt the sincerity of the individual legislator drafting amendments in a committee room at eleven at night. But sincerity of the individual participant has never been the relevant unit of ethnographic analysis. The relevant unit is the institution, and the institution’s behaviour - measured not in intentions but in the allocation of its scarce commodity, legislative time - tells a somewhat different story than the one told in the press release.
What the calendar shows is that “debating” has itself become the deliverable. A regulatory body that produces continuous debate, rather than continuous rule, occupies a peculiar structural position: it satisfies the appearance of governance - the hearings, the stakeholder consultations, the position papers commissioned from institutes whose funding arrives, discreetly, from the very firms under discussion - while deferring the moment at which any actual constraint might bind on a company’s balance sheet. The debate is not the obstacle to regulation. In a considerable number of cases the debate is the regulation, in the sense that it is the entire observable output, the potlatch rather than the harvest.
The revolving door here does not require a single named villain, which is precisely what makes it interesting to the field anthropologist rather than the prosecutor. One need only trace the ordinary career path: policy official drafts framework, framework generates a market for compliance consultancy, official’s expertise in the framework becomes commercially valuable to the firms the framework was meant to constrain, official departs public service for private retainer, framework is revised with the benefit of that same expertise now supplied from the other side of the table. No memo need pass between these parties. The incentive structure performs the coordination that conspiracy would otherwise require, and it performs it with more reliability than any conspiracy ever managed, because it does not depend on anyone’s discretion.
Consider the stakeholder consultation as a specific ceremonial object, comparable to the potlatch feast in that its value lies in the conspicuous expenditure of institutional attention rather than in any transfer of decisive resource. A consultation that receives contributions from hundreds of interested parties and results in a document reflecting the concerns of the best-resourced dozen has not failed at its stated function; it has succeeded at its actual one, which is to confer upon the eventual policy the appearance of having been arrived at by exhaustive deliberation rather than by the ordinary weight of lobbying capacity. The forty-seven-page consultation summary, one might note, invariably outlasts in circulation any enforcement action that follows from it.
None of this requires malice on the part of Europe’s lawmakers, most of whom would describe their August sessions as urgent and their motives as protective of the citizen. The anthropologist does not dispute the motive. The anthropologist observes only that an institution’s actual function is most reliably inferred from what it spends its time producing, and that a continent-wide deliberative apparatus which has, across several seasons, generated an abundance of hearings and a comparative scarcity of binding constraint on the firms whose products occasion the hearings, is an apparatus optimised for something other than the constraint.
A visitor unfamiliar with this particular civilisation, shown only the record of activity - the recurring sessions, the recess and its faithful resumption, the stakeholder papers filed and answered and refiled - would conclude that the institution’s actual product is the appearance of vigilance, produced continuously, consumed by an anxious public, and renewed each August precisely on schedule.