or What Argentina's AI Corporation Law Might Get Right, By Accident...
Argentina's president just proposed a legal category for corporations with no human owners, no human directors, and no human accountability.
Javier Milei published an opinion piece in the Financial Times last week, laying out three pillars of new legislation his government has submitted to Congress.
Pillar one: wholly unregulated AI.
Pillar two: a new corporate category in Argentine law called "non-human corporations": entities operated entirely by AI agents, where human shareholders "may participate, but are not required."
Pillar three: a low corporate tax rate to make Argentina the destination of choice.
He opened his pitch by celebrating the Dutch East India Company. His vision is for Buenos Aires to become "what Amsterdam was for the age of sail."
Unfortunately, the Amsterdam he has in mind is the one that, via the 1602 founded Dutch East India Company, spent two centuries committing massacres, running colonial monopolies, and trading hundreds of thousands of enslaved people. Inspirational stuff.
Your immediate reaction to "AI-run corporation with no human owners required" may be dystopian. That would be fair. A corporation is already a legal excuse to extract profit while limiting personal liability.
Remove the human element entirely and we might be left with extraction without even the pretence of any conscience.
Milei's framing doesn't help change this. "We are open for business," he says. "Keep AI unregulated." Whilst his go-to analogy is a colonial trading empire that operated by force and monopoly for two centuries. So yes. The concern that AI-run corporations become accountability-free extraction machines is entirely reasonable.
There are, however, two sides to every coin.
If he create the legal structure for an AI-run organisation to extract value from people and the planet with no human owner to hold responsible, with super attractive tax rates at that, that very same legal structure will exist for the opposite.
Someone could set up an autonomous organisation whose sole mandate is to protect people. To hold land in perpetuity, the way conservation trusts do across millions of acres, until the funding dries up or the board changes direction. To distribute wealth rather than accumulate it. To monitor corporate behaviour and intervene. To do the things that environmental and social organisations currently struggle to do because they depend on human decisions, donor goodwill, and political winds that shift every four years.
A legal entity with no human shareholders to satisfy, run by AI agents, pointed at preservation rather than profit. Think of the possibilities.
Milei picked the right analogy for the wrong reason. People eventually, imperfectly, partially dismantled the colonial era's extraction machinery by building different kinds of institutions with different purposes. If AI-run corporations gain legal standing, the same opportunity exists. Build the version that protects rather than extracts. Use the same legal tool. Intention not reinvention.
Milei wants Argentina to be old Amsterdam. That’s his choice. It’s up to us to think about how to build more like new Amsterdam, using his generosity.
If a fully autonomous AI organisation existed solely to protect people and/or the planet, with no shareholders, no donors, and no political masters, what should we task it with first?