Antonio Tejeda Encinas | CEO of META Channel Corporation | President of the Euro-American Committee of Digital Law – CEA Digital Law<\/p>\n
The debate on artificial intelligence tends to run aground on a badly framed question: whether machines should have legal personality. It is a question that makes headlines (it already did in 2017, when the European Parliament floated the idea of an \u201celectronic personality\u201d and legal scholarship reacted almost unanimously against it), but it does not solve the real problem. The issue is not whether an AI can be a \u201cperson\u201d, but how an economic activity should be legally organised when some of its decisions are no longer taken by natural persons, directors or employees, but by autonomous agents capable of executing instructions, entering into contracts, moving information, triggering payments and coordinating processes within complex digital systems.<\/p>\n
To address that problem, we use the term ALE (Autonomous Legal Entity<\/em>) here as a functional category of analysis. The expression already circulates in debates on autonomous organisations, DAOs, artificial intelligence agents and algorithmic structures, but it is not yet a settled legal form or a fully stabilised doctrinal category. Its value lies in bringing within a single framework a question the law will have to confront: how to fit autonomous agents, algorithmic systems or decision-making infrastructures within a structure that is recognisable, answerable with its own assets and legally governable.<\/p>\n The key word is container. A traditional company already is one. A limited company, an LLC, a foundation, a trust or an SPV are all forms that make it possible to ring-fence assets, allocate responsibilities, enter into contracts, receive income and operate under governance rules. The difference is that those moulds were designed for human organisations (shareholders, directors, authorised representatives, investors). The ALE raises something else: what happens when the operational core is no longer solely human and becomes, in part, algorithmic.<\/p>\n This is not about a company that \u201cuses AI\u201d: that already exists and raises no new problem. Any company relies on AI to issue invoices, analyse risks or manage documents. The problem arises when the agents are not assisting a person but carrying out a substantial part of the business itself: negotiating parameters, triggering smart contracts, cross-referencing data, ordering payments or coordinating other agents within predefined limits. If the system acts, who is answerable? The founder, the model provider, the developer of the agent, the company that deployed it, the operator with supervisory capacity, or the vehicle itself in whose name it was acting. Without a clear answer, technical autonomy becomes legal unaccountability. That is the underlying risk.<\/p>\n That is why an ALE is not a grant of freedom to machines, but a technique for imposing legal discipline on technological autonomy. Its purpose is not to free software from the law, but to stop it operating in a vacuum of attribution: to identify an operational unit, endow it with assets, set internal rules, define powers, record decisions, audit actions, provide for human oversight and determine liability towards third parties.<\/p>\n It is worth placing the term between two existing positions that point in opposite directions.<\/p>\n Shawn Bayern, a legal scholar and software developer, advances a deflationary thesis: existing company law already allows de facto legal personality to be conferred on an autonomous system, without new legislation, by means of a memberless LLC (zero-member LLC<\/em>) in certain US states. His message, in essence, is that this is already possible and is not that unsettling. At the other extreme, Lynn LoPucki, in Algorithmic Entities<\/em>, warns of precisely the opposite: entities controlled by algorithms are already viable, carry serious risks and should be discouraged rather than facilitated.<\/p>\n The reading we put forward here does not sit comfortably at either extreme. It assumes neither that the phenomenon is impossible nor that it should be left to run its course; nor does it assume that prohibiting it would suffice. It starts from the premise that these arrangements are going to happen and offers the only legally serious response: that they should happen within a demanding container.<\/p>\n A common confusion between these vehicles also needs to be cleared up:<\/p>\n A US LLC is a limited liability company: a classic, flexible mould, built on assets and contract. A DAO, by contrast, is not necessarily a legal person but a form of decentralised organisation based on digital rules, smart contracts<\/em>, tokens<\/em> and distributed governance mechanisms. A DAO can exist as a technology community without clear legal personality, and then the age-old problems resurface: who signs, who is liable, who sues or is sued, who holds the assets in custody and which law governs the structure.<\/p>\n Wyoming has tried to solve part of this. Its DAO LLC<\/em> (2021) wraps a DAO inside an LLC, bringing algorithmic governance into a recognised vehicle: the DAO remains the form of governance, the LLC is the container, and the DAO LLC is the bridge between the two. It is not a revolution in company law, but the adaptation of a familiar mould. The DUNA<\/em> (2024) follows a different logic: a decentralised association that does not distribute profits among its members (although it may carry on economic activity, enter into contracts and open accounts), designed for blockchain networks and token<\/em> projects, with a minimum membership threshold below which it ceases to be \u201cdecentralised\u201d. It matters because it shows that the discussion no longer revolves solely around commercial companies, but also around communities, protocols and collectively held assets.<\/p>\n The ALE is broader than any of these vehicles. It may make use of an LLC, a DAO LLC, a foundation, an SPV, a trust or a form created ad hoc<\/em> by the legislature. It is defined not by its corporate type but by its function: giving legal structure to an autonomous or semi-autonomous activity carried out by software. Not every DAO is an ALE, and nor is every LLC that uses AI. But an ALE can combine corporate, contractual, fiduciary and technological techniques to make an autonomous activity governable. The DUNA, geared as it is towards communities of holders, is a poor fit as a vehicle for an agent operating as a single unit: it serves to illustrate the family of problems, not as a reference model.<\/p>\n Europe has taken a different route. The EU Artificial Intelligence Act (AI Act) does not create autonomous entities or grant legal personality to AI systems. Its architecture is built on roles (provider, deployer, importer, distributor, authorised representative, product manufacturer), supplemented by a specific layer of obligations for general-purpose AI (GPAI) models, following a logic closer to product safety than to company law. It is a regulation of obligations, risks, transparency and human oversight, and as such it is a prudent one: it regulates the chain of responsibility for the system well.<\/p>\n But that logic does not resolve the corporate question. The AI Act answers the question of who must comply with which obligations when an AI system is developed, placed on the market or used. It does not answer how to structure an economic unit in which several agents operate in coordination, manage assets and generate legal consequences on an ongoing basis.<\/p>\n And here lies the obstacle that is usually left out. The underlying reason why Europe cannot simply replicate the Wyoming model is not merely philosophical: it is a matter of company law. Under Spanish law, even when the director of a capital company (sociedad de capital, i.e. a public or private limited company) is a legal person, a natural person must be appointed to perform the duties of the office on a permanent basis. That requirement reveals an idea common to continental company law: ultimate management of a company remains anchored in identifiable human beings. The same can be seen, with its own nuances, in other European legal systems, where companies without members or without a human governing body do not fit easily. The consequence is uncomfortable and should be faced head-on: a European ALE would not be built by adding an article to the AI Act. It would reach into the core of company law. That is an undertaking of a different order of magnitude, and any serious proposal must start there.<\/p>\n Once the obstacle is acknowledged, the thesis stands up better, not worse. If Europe wants to preserve its standard of accountability and legal certainty, it would do well to anticipate the problem, because the alternative to a suitable container is not traditional human control left intact: it is a scattering of automated activities with fragmented liability, contractual chains that are hard to reconstruct, and agents connected to multiple tools with no clear architecture of attribution.<\/p>\n A well-designed European ALE would be the opposite of an opaque entity. It would be an entity of reinforced control: identification of the autonomous systems acting on its behalf, a record of relevant decisions, limits on what they may do, rules for human intervention, technical audit, compulsory insurance in certain sectors, minimum capital, a legally responsible person, traceability of smart contracts<\/em>, a regime for updating agents, suspension mechanisms and reporting obligations. The container does not dilute liability; it makes it traceable and enforceable.<\/p>\n This is not academic speculation. Agentic AI is shifting the debate from the tool to the operation itself. Using a model to draft, classify or summarise is one thing; deploying agents that query databases, execute orders, interact with customers, negotiate terms, consume APIs and leave a real legal and economic footprint is quite another. Once that line is crossed, an AI use policy is no longer enough: what is needed is an agent governance architecture and, in more advanced set-ups, dedicated vehicles to encapsulate risks, ring-fence assets and document actions.<\/p>\n The relevant question will not be whether ALEs should exist. They will exist in one form or another, even if they are not called that: special purpose vehicles with built-in agents, DAOs wrapped in recognised entities, automated fiduciary structures, funds managed by algorithmic rules. The question is whether those structures are designed well or badly. Badly designed, they are screens for unaccountability and tools for regulatory arbitrage. Well designed, they are the piece that keeps the autonomous economy within an auditable and demanding framework.<\/p>\n Corporate law was born to separate assets, limit risk and organise the governance of complex activities. It now has an additional task: to make artificial autonomy legally governable. The aim is not to grant rights to AI; it is to prevent it from operating without attributable duties. And that does not mean emptying companies of people, but properly articulating human intervention, technical autonomy and liability backed by assets. That is the next layer of digital corporate law, and it would be wise to start building it before we need it.<\/p>\n A note on the term and the sources.<\/strong> \u201cALE\u201d (Autonomous Legal Entity<\/em>) is used here as a functional category of analysis, not as a legal form recognised in positive law or as a settled doctrinal concept. The article connects it with existing debates on autonomous organizations<\/em>, algorithmic entities<\/em>, zero-member LLCs<\/em>, DAO LLCs<\/em>, DUNAs<\/em> and structures of algorithmic governance. Core references: the EU AI Act, which organises responsibility around the roles of operators rather than any legal personality of AI; Wyoming legislation on the DAO LLC (2021) and the DUNA (2024); Shawn Bayern\u2019s work on autonomous organisations and the zero-member LLC<\/em>; Lynn LoPucki\u2019s critique in Algorithmic Entities<\/em>; and Daniel H\u00e4usermann\u2019s response from the perspective of continental and Swiss law.<\/p>\n","protected":false},"excerpt":{"rendered":" Antonio Tejeda Encinas | CEO of META Channel Corporation | President of the Euro-American Committee of Digital Law – CEA Digital Law The debate on artificial intelligence tends to run aground on a badly framed question: whether machines should have legal personality. It is a question that makes headlines (it already did in 2017, when […]<\/p>\n","protected":false},"author":1,"featured_media":40592,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[3413,3409,3408],"tags":[3441,3597,3633,3740,3763,3428],"class_list":["post-40374","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-direito-societario-e-contencioso","category-inteligencia-artificial","category-regulacao-e-compliance","tag-advocacia-e-tecnologia","tag-antonio-tejeda-encinas","tag-arquitetura-de-conformidade-regulatoria","tag-arquitetura-juridica","tag-ativos-digitais","tag-blockchain"],"_links":{"self":[{"href":"https:\/\/metachannelcorp.com\/pt\/wp-json\/wp\/v2\/posts\/40374","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/metachannelcorp.com\/pt\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/metachannelcorp.com\/pt\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/metachannelcorp.com\/pt\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/metachannelcorp.com\/pt\/wp-json\/wp\/v2\/comments?post=40374"}],"version-history":[{"count":0,"href":"https:\/\/metachannelcorp.com\/pt\/wp-json\/wp\/v2\/posts\/40374\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/metachannelcorp.com\/pt\/wp-json\/wp\/v2\/media\/40592"}],"wp:attachment":[{"href":"https:\/\/metachannelcorp.com\/pt\/wp-json\/wp\/v2\/media?parent=40374"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/metachannelcorp.com\/pt\/wp-json\/wp\/v2\/categories?post=40374"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/metachannelcorp.com\/pt\/wp-json\/wp\/v2\/tags?post=40374"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}Two poles in the literature<\/h5>\n
LLC, DAO, DAO LLC, DUNA<\/h5>\n
The European path and its real obstacle<\/h5>\n
The container does not remove liability: it organises it<\/h5>\n
What does this mean for businesses?<\/h5>\n