On May 29, 2026, Argentina's executive branch sent Congress a bill to replace the 54-year-old General Companies Law (Law 19,550) with a framework built for firms that run on algorithms rather than payrolls. Buried in Article 14 is the sociedad automatizada — a company that pursues its business purpose "through autonomous algorithmic systems or artificial intelligence agents, without requiring dependent workers or human resources for its ordinary operation," according to the government's own summary on Argentina.gob.ar. The Senate's General Legislation Commission opened formal debate on June 24, 2026, with Deregulation Minister Federico Sturzenegger presenting the bill alongside Inspector General Alejandro Ramírez, per the Senate's own press office.
This is the legal chassis for the vision Milei laid out in a June 4 Financial Times op-ed, co-written with Sturzenegger: a three-pillar strategy of unregulated AI development, the new automated-company category, and a competitive corporate tax regime to match, as reported by PYMNTS. Milei framed it in explicitly historical terms — "let Buenos Aires be for AI what Amsterdam was for the age of sail" — casting limited liability itself as the innovation the AI era now demands, the way it was for the Dutch East India Company in 1602.
The mechanics, not the marketing
The bill does not create a company with no human in it. What's automated is ordinary operation — the routine execution of a business, not its governance. Management stays with human administrators, and the drafted text calls for a designated human responsible party with "explicit obligations of supervision and effective responsibility before the State and third parties." The automated company "will respond with its assets" for damage its algorithmic systems cause to third parties. That's a real accountability structure, not a legal loophole dressed up as one — and it's a meaningfully more cautious design than the "non-human corporation" framing that dominated English-language coverage of Milei's op-ed.
The steelman: Harari's asymmetry argument deserves a real answer
Historian Yuval Noah Harari published a rebuttal in the same paper, and his core point is worth taking seriously rather than dismissing as technopanic. Human executives face prison for fraud and negligence; an AI system cannot be jailed, so a company that genuinely runs on autonomous agents could take on more risk than a human-run firm would, precisely because its nominal operator faces a categorically weaker deterrent. Harari's sharper warning — that granting legal personhood to AI-run entities risks producing "not a company-state, but an AI-state" — is the strongest version of the case against the bill, and Milei's own reply (that bankruptcy, asset seizure, and forced liquidation are real deterrents even without a body to jail) doesn't fully answer it, because those penalties fall on the company's assets, not on whoever chose to deploy the system in the first place.
That gap is exactly where Argentine legal critics have focused. As the Buenos Aires policy outlet Cenital reports, digital-rights lawyer Javier Pallero flags that Article 14's cap on liability to corporate assets cuts against long-standing Argentine doctrine that lets fraud victims pierce the corporate veil and sue owners personally — and that "algorithmic randomness" could become a ready-made excuse for decisions that were, in fact, human choices about training data, deployment timing, and task scope. The Argentine Chamber of Commerce raised a blunter version of the same question during committee testimony: who answers for criminal fraud if the automated company's treasury is empty by the time a claim is filed?
Why the bill is still the right instinct
None of this is a reason to block the reform. Argentina's current companies law long predates the internet, and forcing an AI-operated logistics platform or DAO-style venture into a legal shell designed for factories is its own kind of distortion — one that pushes real economic activity into informal or offshore structures with even less accountability, not more. The bill pairs the automated-company category with a broader push, reported by We Are Innovation, to collapse Argentina's more than 165 overlapping taxes into a single 15% flat rate — a genuine attempt to make the country legible to capital that currently avoids it, not merely a tax giveaway to AI firms specifically.
The honest test is what happens between now and floor votes. The bill is still in Senate committee; nothing described here is enacted law. The fix Harari's critique points toward isn't scrapping the automated-company concept — it's hardening Article 14's supervisory-liability language so the named human administrator carries personal exposure proportionate to the autonomy they've granted the system, closing exactly the loophole Pallero and the Chamber of Commerce identified. A regime that makes the human name behind the algorithm legible to regulators and creditors is a legitimate innovation. One that lets that name disappear behind a depleted balance sheet is not — and right now, the text leaves room for both readings.