On June 4, 2026, Argentine President Javier Milei and Deregulation Minister Federico Sturzenegger published a Financial Times op-ed, "Argentina invites AI to free itself," promising to let artificial intelligence develop "without the deadly hand of premature and poorly understood regulation." Milei closed with a flourish: "Let Buenos Aires become for AI what Amsterdam was for the age of sail." Behind the rhetoric sits real legislation already before Congress. Two of its three pillars are the kind of pro-investment, proportionate policy this publication has long argued for. The third is a legal fiction that should not survive committee.
The fiscal pillar is sound
The centrepiece is the "Súper RIGI," sent to the Chamber of Deputies in late May 2026 as Message N°181/2026, signed by Milei, Cabinet Chief Manuel Adorni and Economy Minister Luis Caputo. It expands Argentina's 2024 large-investment regime to "new industries" — artificial intelligence, semiconductors, data centres, advanced biotechnology, lithium batteries and green hydrogen. Adhering firms pay income tax at 15%, against the 25% of the original RIGI and Argentina's 35% general rate, with dividends taxed at 7% and falling to 3.5% from the fourth year. The state guarantees tax, customs and exchange-rate stability for 30 years. Entry requires a minimum investment of US$1,000 million, with 20% deployed in the first two years.
For a country that has spent decades scaring off capital with inflation, capital controls and policy whiplash, a transparent, durable fiscal compact is a genuine competitive offer. Data centres are mobile; they follow cheap energy, cool climates and legal certainty. Argentina has Patagonian cold, stranded gas at Vaca Muerta and, now, a 30-year stability pledge. None of that requires inventing new legal persons. It requires the boring virtues of predictability — exactly what the Súper RIGI delivers.
The light-touch posture on AI itself is also defensible. The strongest case for moving slowly on hard AI statutes is real: the EU's own AI Act is already being reopened and delayed because its risk categories aged badly against generative models. Legislating a fast-moving technology in detail tends to lock in yesterday's assumptions. A jurisdiction that waits, watches and applies existing consumer-protection, liability and data law to concrete harms is not being reckless — it is being humble about what regulators can foresee. On this, Milei has a point.
The 'non-human corporation' is where it breaks
The second pillar abandons that humility. A parallel Sturzenegger bill rewrites company law to create the "Sociedad Automatizada" — an entity that "operates through algorithms or AI without requiring employees," alongside blockchain-based DAOs. Both would receive "full legal personality and limited liability," and human shareholders would be optional, not required. This is the "non-human corporation" Milei pitched in the FT: a company that can own assets, sign contracts and shield its operators from liability, run by software with, potentially, no natural person behind it.
Steelman it first. The bill's drafters argue that when AI systems "exercise independent judgment in unpredictable environments, limited liability is not a luxury… it is a condition of possibility for their existence." There is a coherent intuition here: autonomous agents that transact at machine speed need a legal wrapper, and forcing every micro-transaction through a human signatory is friction that kills the use case. Wyoming's DAO LLC statute and the Marshall Islands' DAO law reflect the same impulse.
But those precedents kept a human or a registered member ultimately answerable. Milei's version does not, and that is the fatal flaw. As Buenos Aires technology-law specialist Alan Daitch put it, the proposal "does not close legally." Argentina's Ley General de Sociedades requires that shareholders be persons: "Someone puts capital and someone is responsible if something goes wrong." AI, Daitch notes, "is not a subject of law" — it cannot be sued, fined, jailed or made to pay damages. "At bottom, there is always a physical or legal person who responds, even if management is automated." His verdict is that the category offers "no structural novelty; only new clothing for figures that already exist."
This is not a pro-regulation complaint. It is an accountability complaint, and accountability is what makes markets work. Limited liability is a bargain: society grants it in exchange for a known owner who bears residual responsibility and whose identity supports enforcement against fraud, money-laundering and tort. Strip out the human and you do not deregulate — you create a liability black hole that launderers, sanctions-evaders and fraudsters will fill faster than any AI lab. A genuinely pro-innovation framework would let AI agents act on behalf of accountable principals, not replace them.
The rest of the world is hedging the other way
Argentina's gambit also lands against a strong counter-current. One day earlier, on June 3, 2026, the European Commission unveiled its Technological Sovereignty Package — a Cloud and AI Development Act and Chips Act 2.0 — to reduce dependence on US infrastructure, and Canada launched an "AI for All" strategy built on the same sovereignty logic. Those approaches carry their own risk of protectionist overreach. But they share a premise Milei discards: that AI sits on infrastructure and institutions someone must remain answerable for.
The smart version of Milei's plan is achievable. Keep the Súper RIGI's fiscal certainty. Keep the restraint on premature AI statutes. Drop the personhood fiction and instead modernise agency and liability law so AI agents can transact efficiently under a named, accountable owner. That is proportionate, pro-innovation regulation. Granting legal personhood to a machine with no human behind it is neither.