On May 29, 2026, Argentina's executive branch submitted to Congress a comprehensive reform of the country's corporate law—a draft General Companies Law that, if passed, would repeal Law 19.550, in force since 1972. Among its nearly 295 articles sits something genuinely unprecedented: a new category called the Sociedad Automatizada, a company that operates entirely through autonomous algorithmic systems or artificial intelligence agents, with no mandatory human employees for ordinary operations and, crucially, no requirement for human shareholders at all.
Deregulation Minister Federico Sturzenegger, who co-authored the framework with President Javier Milei and defended it before the Senate's General Legislation Commission on June 24, has framed this as an economic opportunity comparable to the 17th-century invention of limited liability. "For a country with pretensions to lead, it is absolutely imperative to incorporate these technologies," Sturzenegger told senators. In a June 3 op-ed in the Financial Times co-authored with Milei, the government declared: "Let Buenos Aires be for AI what Amsterdam was for the age of sail."
The Strongest Case For It
Before dismissing this as libertarian grandstanding, the core legal problem the bill addresses is real. Argentina's 1972 corporate law was designed for a brick-and-mortar industrial economy that predated the public internet by two decades. Globally, AI-operated entities fall into dangerous legal grey zones: a 2023 U.S. federal court ruling, Sarcuni v. bZx DAO, found that a blockchain-based autonomous organization operated as a general partnership, stripping members of liability protection entirely. Without a recognised legal category, AI-operated firms face a binary choice between regulatory limbo and exposure that deters legitimate investment.
Studying Ireland's transformation—Sturzenegger cited how the country grew its per-capita income from roughly 60 percent to 115 percent of the EU average between the 1980s and 2000s partly through targeted regulatory attractiveness—is not absurd. Argentina's economy grew 4.4 percent in 2025 according to INDEC, reversing a 1.3 percent contraction in 2024, and the government is searching for structural anchors for that recovery. A competitive corporate framework for the AI era is a plausible one.
What the Law Actually Creates
Under Article 14 of the draft, any company type permitted under the new law may qualify as a Sociedad Automatizada by meeting two conditions: the automated character must be declared expressly in the bylaws, and the word "Automatizada" must appear in the company name so that third parties know immediately they are dealing with a non-human-operated entity. The bill also introduces Sociedades Descentralizadas Autónomas Operativas (DAOs)—entities structured through decentralised blockchain protocols that can issue participations as digital tokens.
On liability, the draft is specific: the automated company "responds with its assets to third parties for damages caused by its autonomous algorithmic systems or artificial intelligence agents." This is corporate-asset liability, equivalent to what a conventional limited-liability company already carries. Crucially, the bill also requires that DAOs and automated companies identify their ultimate beneficial owners to comply with Argentina's Financial Information Unit (UIF) and FATF standards—anonymous operations are explicitly excluded.
The Accountability Gap That Remains
Historian Yuval Noah Harari published a direct counter-argument in the Financial Times on June 8, 2026, under the headline "We must not grant AI agents legal personhood." His central point is precise: "The maximum sanction that deters human executives—prison—is irrelevant to AI." A human director who authorises fraud fears consequences that an autonomous algorithmic system simply does not register. Harari warned that countries adopting such frameworks risk becoming "an AI state" rather than a company-state, potentially governed by non-human entities pursuing optimisation objectives with no stake in human welfare.
Argentine legislators raised parallel concerns. Senator Jorge Capitanich, speaking during the June 24 session, warned that responsibility cannot remain in a "caja negra" (black box). He demanded clear identification of beneficial owners, human-reviewable decisions, and legally imputable representatives. Senator Martín Soria invoked the notorious Simplified Joint-Stock Companies (SAS) introduced under the Macri administration, several of which were exploited by criminal organisations—a reminder that novel corporate forms can be weaponised faster than oversight can adapt.
These are not frivolous concerns. The bill's asset-liability rule is adequate for ordinary commercial disputes but underpowered for harms that exceed a company's balance sheet: algorithmic discrimination at scale, data breaches affecting millions of users, or an autonomous trading agent that amplifies a financial crisis. In these scenarios, knowing which AI system caused harm and who designed its objectives is essential—and the bill currently provides no mandatory audit trail or human oversight mechanism beyond beneficial-owner registration.
A Proportionate Path Forward
The bill's transparency requirements—mandatory name disclosure and beneficial-owner identification—are genuinely meaningful starting points, and dismissing them as cosmetic would be unfair. What the framework still lacks is a mandatory human legal representative requirement for litigation purposes, a minimum algorithmic-impact disclosure standard for automated companies operating at scale, and a regulatory sandbox mechanism that would let authorities monitor novel entities before they are granted unlimited commercial scope.
These additions need not defeat the bill's competitive logic. Wyoming's 2021 DAO LLC law, the closest U.S. analogue, similarly granted DAOs limited liability but required a registered agent—a single human or entity accountable for legal process. Argentina could adopt comparable rules without abandoning the three-pillar strategy of minimal regulation, new corporate categories, and competitive taxation.
What makes Argentina's experiment genuinely important is not the ideology behind it—comparable debates have occurred in every jurisdiction grappling with autonomous systems—but the fact that it is the first national attempt to codify AI legal personhood in general corporate law rather than a sandbox or sector-specific carve-out. The EU AI Act, finalised in 2024, took the opposite approach: obligations cascade through providers, deployers, and operators, always anchoring liability on a human chain. Argentina's bill bets that clarity of legal personhood for the AI entity itself produces better outcomes than distributed human accountability chains.
That bet may be right. But right now the bill is a bold first draft, not a finished answer. The Senate debate that Sturzenegger opened on June 24 is the right place to close the gap.