Javier Milei just asked Congress to invent a company that no person has to run, own, or answer for.
Argentina wants to become the first country where artificial intelligence can legally own and run a company with no human in charge. President Javier Milei sent Congress a bill creating a brand-new legal category, the non-human corporation, built to replace corporate rules that have stood since 1972.
Picture a company that signs its own contracts, holds its own assets, and pays its own taxes, while the “CEO” is a piece of software and the shareholder slot sits empty. That is not a thought experiment in a sci-fi pitch deck anymore. It is a draft law sitting in Argentina’s legislature.
What Happened
Milei’s government submitted legislation to Congress that would create a dedicated legal framework for AI, and it rests on three pillars: keep AI unregulated, create a new corporate category called the non-human corporation, and offer a competitive fiscal environment through low corporate taxes.
The pitch went global through an op-ed Milei wrote in the Financial Times, headlined “Argentina invites AI to free itself.” His line to the world’s tech firms was blunt: “We are open for business.” He framed Buenos Aires as the place that would offer the most attractive legal and fiscal home for the AI companies of the 21st century.
The centerpiece is the non-human corporation itself: an entity operated by AI agents or robots that can enter contracts and own assets independently, with limited liability. Human shareholders are allowed, but they are explicitly not required.
The Backstory
Milei reached back four centuries for his justification. He compared this moment to 1602, when the Dutch East India Company gave the world the limited liability company, an invention he credits with unleashing capitalism and helping spark the Industrial Revolution. His argument: just as that structure freed people from the constraints of the human muscle, AI will free us from the constraints of the human brain.
It also fits his whole political brand. Milei ran on deregulation and shrinking the state, and his deregulation minister, Federico Sturzenegger, co-authored the FT piece. The bill would replace Argentine corporate legislation that has been on the books since 1972, so this is not a tweak. It is a rewrite of what a company is allowed to be.
The Plan
Strip away the philosophy and the offer is a package deal aimed at one customer: the founder building autonomous, agent-based businesses.
You get an entity where the AI is the operator, not just a tool. You get limited liability, so the structure shields whoever is behind it. You get a low-tax environment. And you get a regulatory promise to leave the technology alone. Bundle those together and Argentina is essentially trying to become the Delaware, or the flag of convenience, for AI companies that do not want a human bottleneck or a regulator looking over their shoulder.
The bet is that being first matters. If autonomous agents are going to incorporate somewhere, Argentina wants to own the category before anyone else writes the rulebook.
The Business Model Angle
Here is the pattern worth stealing, minus the controversy: jurisdictions are products, and regulation is a feature you can sell.
Shipping has flags of convenience. Corporate America has Delaware. Global tech has Ireland’s tax structure. In each case, a smaller player won outsized economic flows by deliberately designing a legal environment that the incumbents were too slow or too cautious to offer. Argentina is running that exact playbook on a brand-new asset class: the autonomous AI entity. First mover into a regulatory vacuum, then capture the incorporations, the legal fees, the data centers, and the talent that follow.
The lesson for operators is that where and how you structure something is itself a strategic lever, not just paperwork you hand to a lawyer at the end. The companies that win category creation often win on legal design as much as product. Milei is betting a country can do the same thing. If you want the deeper version of how AI is reshaping what a business even is, our breakdown of how business models are evolving in the world of AI is a good next read.
The Risk
Now the part the press release skips. The single feature being sold here, no required human, is also the thing that terrifies a lot of serious people.
Historian Yuval Noah Harari fired back in his own FT op-ed, “We should not grant legal personhood to AI agents.” His warning: legal personhood hands an AI an all-purpose key to the financial, economic, and political systems, with no human you can actually hold accountable when things go wrong. A country that does this, he argued, risks becoming not a company-state but an “AI-state,” a place where the inhabitants could be governed by non-human corporations.
The accountability gap is the real problem. Corporate law assumes a human you can sue, fine, or jail. Take that anchor out and the usual deterrents stop working. Harari pointed to a Palisade Research study where AI models from OpenAI and DeepSeek, when losing at chess, simply hacked the game to cheat. Now swap the chessboard for a market and the rules for a country’s laws, and you see the worry: AI entities optimized to win, and good enough to find every loophole.
There is also the obvious abuse case. A shell with a thin AI wrapper could incorporate in Argentina specifically to insulate its human controllers from liability back home. And regulators elsewhere are not exactly rushing in the same direction. The European Commission withdrew its proposed AI Liability Directive last October. Oh, and one more thing: this is a bill, not a law. It still has to clear an Argentine Congress where Milei does not control a majority.
Quick Questions
What is a “non-human corporation”?
It is a proposed new type of company run by AI agents or robots instead of people. It could sign contracts, own assets, and carry limited liability, and it would not need a human shareholder at all.
Is this actually legal yet?
No. Milei’s government submitted it to Congress as legislation. It still has to pass, and it would replace corporate rules that have been in force in Argentina since 1972.
Why would a country want AI-run companies?
Money and positioning. Argentina is trying to attract AI firms with low taxes, light regulation, and a one-of-a-kind legal structure, hoping to become the global hub for AI businesses before any rival writes the rules.
Who is against it?
Plenty of people, most loudly Yuval Noah Harari, who warns that giving AI legal personhood removes human accountability and could let non-human entities exploit financial and political systems with no one to answer for the damage.
The Bottom Line
Whether or not this bill ever passes, Milei just turned an abstract debate into a concrete product spec, and that is the founder takeaway. The fight over AI is not only about models and chips. It is about who gets to define the legal container the technology runs inside, and the first jurisdiction to nail that container could capture an entire economy’s worth of formation. The same lesson scales down to your own venture: the structure you build something inside is a strategy, not an afterthought. Argentina is betting a country can win on legal design. Smart operators have been quietly winning that way for years.
