Europe debates AI regulation to manage global risks
This matters for managing the risks and reaping the benefits of artificial intelligence, affecting societies and economies globally.
We are told that law and capability are natural enemies, that whatever a parliament writes must subtract from what an engineer may build. Europe’s lawmakers, occupied again this 25th of August 2026 with the machinery of artificial intelligence regulation, are cast by most commentary as referees standing between a reckless technology and a frightened public. But notice the instrument that has actually entered this debate: not a new model, but a new apparatus of testing - the conformity assessment and the regulatory sandbox now written into the continent’s AI law, by which a system’s claimed capability must be demonstrated against a named risk before it is released onto a market of several hundred million people. What this instrument makes possible, for the first time in this technology’s short life, is the conversion of a philosophical anxiety into an experimental protocol. The question worth asking on this date is not whether Europe favours caution over commerce, but whether its new instrument for testing AI systems actually tests anything, or merely certifies that the paperwork exists.
The story treats Europe’s lawmakers as referees of someone else’s invention - as if the artificial intelligence being debated is the only made thing in the room. But there are two creations here, and the second is the one nobody is watching. The law itself is a made thing. A statute, once passed, goes on acting in a world its drafters cannot fully model: it gets applied by tired officials in twenty-seven capitals, tested by lawyers hunting its seams, stretched by companies who read its thresholds the way a locksmith reads a door. The debate happening in Europe as of 25 August 2026 is not simply about what artificial intelligence will do next. It is about what the regulation will do next, once the debating chamber empties and the text is left to fend for itself in the world.
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 principle operating here, stated plainly, is: regulate the artificial intelligence that others will build, deploy, and profit from, so that its risks are borne collectively while its exemptions are reserved for the regulator. Let us ask whether this principle, universalised, produces coherence or contradiction.
Europe’s lawmakers, gathered in the ordinary business of law-making on this 25th of August 2026, are not merely drafting technical schedules and risk categories. They are, whether they know it or not, testing a maxim before the world has finished writing it down. The maxim is this: that a legislature may impose upon private actors constraints on the use of a technology which it declines to impose upon itself, on the ground that its own use of that technology serves ends too important to submit to the general rule. This is not a hypothetical. It is the structure of every draft that carves out a national-security exception, every clause that permits a state ministry to deploy a system that a hospital or a hiring firm could not lawfully touch.
The story frames Europe’s regulatory debate as lawmakers wisely stewarding a dangerous new technology into safety. But look at what is quietly being decided in Brussels and Strasbourg this August of 2026: not merely rules for artificial intelligence, but the terms on which a vast, collectively built body of knowledge - training methods, architectures, evaluation practices, the accumulated craft of thousands of researchers who published, argued, and corrected one another in the open for a decade - gets fenced, and who is handed the key to the gate.
The official account: Europe’s lawmakers are, as of this August, engaged in the serious business of regulating artificial intelligence - weighing risk against benefit, drafting rules that will let societies and economies capture the gains of the technology while guarding against its dangers. This is the dignified version, and it is not false. It is simply incomplete, in the way that saying the Queen opens Parliament is not false but tells you nothing about who wrote the speech.
The story arrives as: Europe’s lawmakers are heavily involved in debating how to regulate artificial intelligence, a sentence built to be read and immediately forgotten, because it describes a process rather than a fact. Debating is not a state that ends. One notices the date attached to it - 25 August 2026 - and the date is the detail the framing wants left at the edge, because the EU’s AI Act stopped being a debate and started being law, in the part that matters most, three weeks earlier. The obligations covering high-risk AI systems came into force on 2 August 2026. The lawmakers photographed “debating” in the third week of August are debating the enforcement of a rule that already has a compliance deadline behind it, not in front of it.
Thorstein Veblen
One observes that the European Commission’s delegation of technical detail to unaccountable committees is described not as a structural necessity, but as a triumph of efficiency - a mechanism by which the dignified act of legislation is purified of the messiness of public scrutiny. The technocratic position argues that this arrangement allows the “efficient work” to proceed in August 2026, insulated from the delays of political debate, thereby ensuring that the law functions as a precise instrument of governance rather than a symbolic gesture. This is a plausible observation of institutional behavior, and I concede that the separation of high-level mandate from low-level specification is indeed a feature of modern bureaucratic enterprise, not merely a bug. However, the technocratic framework mistakes the method of administration for the purpose of the institution, failing to recognize that the very opacity it praises as efficiency is the primary vehicle for the ceremonial retention of status by the regulated class.
The technocratic argument posits that “technical complexity is delegated downward while political credit is retained upward.” This is accurate, but it misidentifies the beneficiary of the arrangement. The delegation is not merely an administrative convenience for the sake of legislative speed; it is a ceremonial ritual designed to shield the industrial interests from the productive demands of the public. When the Commission takes the bow for regulating AI, it performs a ritual of modernization, signaling to the global market that Europe is a serious, orderly participant in the technological sphere. The actual content of that regulation, settled by standards bodies whose names most citizens could not supply, serves a different function: it provides the legal cover for the continuation of existing power structures under the guise of technical compliance. The “interesting questions” of who certifies conformity and what constitutes a high-risk system are not merely technical; they are the precise points where the ceremonial interest of the leisure class - defined here as the corporate and administrative elite - secures its position against the disruptive potential of genuine public accountability.
Consider the parallel of the potlatch among the Kwakwaka’wakw, where wealth is destroyed not to be hoarded, but to be displayed as a marker of status and power. In the European regulatory apparatus, the “destruction” is not of wealth, but of transparency. The complexity of the delegated acts functions as a form of ceremonial waste, a barrier to entry that ensures only those with the requisite resources - legal, technical, and financial - can navigate the regulatory landscape. The standards bodies are not neutral arbiters of technical truth; they are kinship groups within the institutional tribe, bound by shared educational backgrounds, professional networks, and revolving-door incentives. The “negotiation among interested parties” described by the technocrat is not a flaw in the process; it is the process itself. The opacity is not a side effect of complexity; it is the primary product, consumed by the industrial class to maintain their dominance over the productive apparatus.
The technocratic framework assumes that the legislature’s inability to write detail is a weakness to be corrected by delegation. I argue that this inability is a strength to be preserved. The vague mandate allows the political class to claim credit for action without assuming responsibility for outcome. The specific detail, delegated to the committees, allows the industrial class to shape the rules in their favor without public scrutiny. The result is a double performance: the politicians perform governance for the electorate, while the technocrats perform compliance for the corporations. The “efficient mechanism” is efficient only in the sense that it efficiently transfers value from the public sphere to the private ceremonial sphere.
A rational observer, unfamiliar with the civilisation’s custom of claiming efficiency while practicing opacity, would conclude that the institution’s primary function is not the regulation of technology, but the regulation of attention. The law is not designed to control the AI; it is designed to control the perception of control. The standards bodies are not technical committees; they are status-conferring organs, bestowing legitimacy upon those who can afford to participate in their deliberations. The “high-risk system” is not a technical category; it is a ceremonial boundary, drawn to exclude public scrutiny while including corporate innovation. The technocrat sees a streamlined pipeline; the ethnographer sees a potlatch of complexity, where the destruction of transparency is the gift that buys the loyalty of the industrial class.
Walter Bagehot
The official account: The European Parliament is engaged in a sincere, laborious, and substantive deliberation over the governance of artificial intelligence, driven by a genuine desire to manage global risk and secure economic benefit. The machinery: The legislature is performing a seasonal rite of reconvening to allocate its most scarce commodity - legislative time - over a technology whose consequences are still largely theoretical, thereby prioritizing the appearance of oversight over the operational reality of innovation. The gap between these two is not hypocrisy; it is structural necessity. The institution must look busy to maintain confidence, even when the operational output is merely a refinement of previous cycles.
Your field notes identify a curious seasonal migration: lawmakers dispersing to their national capitals and reconvening in Brussels to resume the annual rite of AI governance. You argue that this deliberation has been conducted for several successive legislative seasons, and you reasonably ask what distinguishes this round from the last. You suggest that the distinction may be ceremonial rather than substantive, and that the sincerity of individual legislators drafting amendments at eleven at night is irrelevant to the ethnographic analysis of the institution. I concede your strongest point here: sincerity is indeed the wrong unit of analysis. An institution’s behaviour is measured not in intentions but in the allocation of its resources, and if the resource being allocated is time spent on a problem that has no clear operational solution, the output is necessarily performative.
However, you stop short of identifying the convention that actually governs this situation. You observe the migration and the repetition, but you do not name the mechanism that binds the dignified ceremony to the efficient outcome. The convention is not that the debate is useless; the convention is that the debate is a buffer against uncertainty. In the late nineteenth century, when the Bank of England faced a sudden drain of reserves, it did not announce its panic to the markets. It raised the discount rate, quietly and decisively, while maintaining the public face of calm. The ceremony of calm was not a lie; it was the mechanism that prevented a run on the bank. Similarly, the European Parliament’s AI deliberations are not a failure of substance; they are a buffer against the market’s impatience. The legislators are not debating the technology so much as they are debating the pace of its integration into society.
Where our frameworks diverge is in the valuation of this ceremonial time. You view the repetition as evidence of ineffectiveness. I view it as evidence of institutional self-preservation. Consider the English Constitution: the Monarch’s ceremonial duties are not mere decoration; they are the anchors of stability. If the Monarch were to perform only the efficient functions of government, the system would lose its dignity, and with it, its legitimacy. The Parliament’s AI debates serve the same function. They allow the technocratic elite to signal caution to the public while allowing the market to proceed with innovation in the shadows. The gap between the dignified version (we are regulating AI) and the efficient version (we are managing public anxiety about AI) is where the real work happens. To demand that the ceremony match the operational reality is to misunderstand the purpose of the ceremony.
You note that the observer arriving without prior knowledge would ask what distinguishes this round from the last. The distinction is subtle but critical: the technology has moved from the realm of speculation to the realm of deployment. In the past, the debate was about if AI should be regulated. Now, the debate is about how much it should be constrained. This shift is operational, not ceremonial. The legislators are not repeating themselves; they are adjusting the dials on a machine that is already running. The convention has shifted from prevention to mitigation. This is a significant evolution in institutional behaviour, even if the press releases remain identical in tone.
The official explanation is admirable. The actual mechanism is rather different, and rather more interesting. The European Parliament is not failing to govern AI; it is governing the perception of AI governance. This is a crucial distinction. If you view the institution as a legislative body that produces laws, you will be disappointed by the slowness and the repetition. If you view it as a confidence-engine that produces stability, you will see the genius of the system. The legislators are not drafting codes; they are drafting reassurances. The code will come later, when the market has stabilized and the public anxiety has been sufficiently dampened by the ceremony of debate.
What the operational analysis reveals that the official account obscures is the timing of regulation. Regulation is not a barrier to innovation; it is a lubricant. By delaying the substantive rules, the Parliament is allowing the industry to mature and the infrastructure to build. When the rules finally come, they will not be a shock; they will be a formalization of what has already happened. The seasonal migration is not a waste of time; it is a necessary period of gestation. The lawmakers are not debating the future; they are negotiating the present. And in that negotiation, the ceremony is not a distraction from the work; it is the work.
The Verdict
Where They Agree
Both debaters anchor their analyses in a common understanding of institutional behavior, observing that the European Parliament’s activities are structured by an unstated convention separating public ceremony from operational reality. For Veblen, this is a “potlatch” that consumes institutional attention to produce the appearance of vigilance. For Bagehot, it is the “dignified” part of the constitution that maintains public confidence while the “efficient” machinery works in the background. This shared premise - that the debate itself is a deliverable - is significant because neither believes the official narrative that the process is primarily about crafting effective regulation. They agree that the institution is, first and foremost, managing perception.
both analysts concur that the technical details of AI regulation, the “codes of practice” and “delegated acts,” are where real power is exercised, and that this exercise occurs away from public legislative scrutiny. They see the delegation of technical complexity to committees and standards bodies as the central mechanism of governance, even as they disagree profoundly on its purpose. This shared focus on the gap between high-level mandate and low-level implementation reveals an unstated consensus: the substance of AI law will be determined by a non-legislative, expert-driven process.
Where They Fundamentally Disagree
The purpose of the ceremonial debate. The empirical disagreement here is over the effect of the performance: does it function as a beneficial “buffer against uncertainty” (Bagehot) or as a “potlatch” that systematically disadvantages public accountability (Veblen)? Bagehot, applying his constitutional framework, argues the empirical effect is stability; the ceremony of debate dampens public anxiety and allows innovation to proceed at a manageable pace, making eventual regulation a “lubricant” rather than a “barrier.” Veblen, from his anthropological viewpoint, sees the empirical effect as the consolidation of power; the ceremony is a “barrier to entry” that ensures only well-resourced corporate and administrative elites can influence the operational rules. The normative disagreement is starker: Bagehot values institutional stability and confidence as necessary social goods, while Veblen views the same stability as a mechanism for perpetuating an inequitable status quo.
The beneficiary of delegated technical authority. The factual core of this dispute is whether the committees that write the technical standards are neutral, expert bodies or “kinship groups” bound by shared interests. Bagehot assumes they are “efficient” mechanisms necessary for governing complexity, their negotiations a normal part of modern regulation. Veblen assumes they are “status-conferring organs” whose primary function is to serve “the industrial class.” Normatively, Bagehot sees delegation as a pragmatic solution to legislative incapacity, a convention worth preserving for functional governance. Veblen sees it as a deliberate strategy to exclude the public and concludes the legislature’s inability to write detail is a “strength to be preserved” because it forces accountability into the open.
The relationship between regulation and innovation. The empirical disagreement is on the sequencing and impact of rules. Bagehot posits that delayed, deliberate regulation allows an industry to mature, and that rules will formalize existing practice rather than shock it - a claim resting on the assumption that the market’s development is a predictable process. Veblen would counter that this delay is not benign gestation but a period where corporate interests entrench themselves, making future regulation largely ceremonial. The normative split is between Bagehot’s belief that managed, stable integration of technology is a public good and Veblen’s belief that such management inherently serves entrenched power at the expense of disruptive, productive change.
Hidden Assumptions
- Thorstein Veblen: Assumes that complexity in regulation functions primarily as a ceremonial barrier to public participation, designed to maintain elite control. If this were false - if, for instance, complexity were an unavoidable byproduct of managing genuinely intricate systems and public participation was facilitated through other means - then his entire critique of the process as a “potlatch” would collapse into a mere complaint about bureaucratic inefficiency.
- Thorstein Veblen: Assumes that the revolving door between regulators and industry creates a uniform class interest that systematically subverts public intent. If this were false - if, for example, personal ethics, professional norms, or countervailing legal pressures routinely prevented regulators from favoring their future employers - then his description of the process as a coordinated “ceremonial retention of status” would be an overstatement.
- Walter Bagehot: Assumes that public confidence in institutions is fragile and depends on the maintenance of dignified ceremonies that obscure complex or potentially alarming operational realities. If this were false - if the public could maintain confidence in institutions that transparently explained their technical challenges and uncertainties - then his defense of the current system as necessary for stability would be moot.
- Walter Bagehot: Assumes that the pace of AI innovation, while fast, is slow enough for deliberative technocratic bodies to eventually produce effective and durable rules. If this were false - if the technology’s evolution is so rapid that any rule is obsolete upon publication - then his entire model of “efficient” behind-the-scenes governance fails, and the ceremony becomes pure distraction.
Confidence vs Evidence
- Walter Bagehot: *The claim that the AI debate has shifted from “if” to “how much” regulation - * the evidence offered is purely observational and circumstantial (the timing of debates). A reader should be suspicious because this is a fundamental claim about legislative intent and the maturity of the technology, yet it is presented as an established fact without reference to specific legislative text or technological milestones that would substantiate the shift.
- Walter Bagehot: *The claim that delayed regulation acts as a “lubricant” for innovation - * this is a central tenet of his argument. The underconfidence here is strategic, masking a strong normative argument with a veneer of caution. The evidence for this is contested economic theory, not a settled fact, and the low tag does not accurately reflect the argument’s load-bearing role.
What This Means For You
When you read about AI regulation, ask a specific question: what is happening in the technical committees? The most important decisions are not on the parliamentary floor but in the groups defining terms like “high-risk.” Be suspicious of coverage that focuses solely on political declarations and ignores the implementation process. Your skepticism should spike when officials claim a law is “landmark” or “comprehensive” before the technical standards are even written. The single most important piece of evidence to demand from any report is the membership list and consultation record of the standards body writing the specific compliance rules. Who is in the room determines what the law will actually do.