On May 29, 2026, Javier Milei’s government sent Argentina’s Congress a bill rewriting the country’s corporate law to introduce a brand-new category: the automated company, run by AI agents or robots with no human employees at all. The plan turned global this week after historian Yuval Harari, author of Sapiens, and Microsoft AI chief Mustafa Suleyman publicly rejected it.
Key Takeaways
- The bill creates a company run by AI agents, with no required human staff.
- Harari and Suleyman warn against granting legal personhood to AI corporations.
- Milei has already answered Harari on X and is preparing a full rebuttal.
What the bill actually says
The bill amends Argentina’s general companies act (Law 19.550) and introduces a new figure called the “automated company”.
Article 14 defines it as a company that pursues its corporate purpose through autonomous algorithmic systems or AI agents, without needing employees or any human resources for day-to-day operations. The automated nature must be written into the bylaws, and the company name must include the word “Automatizada”.
Liability is settled in the same article 14: the company answers with its own assets for damages caused by its systems. The corporation is liable, not the algorithm acting as if it were a person.
In an op-ed co-signed with Federico Sturzenegger, Minister of Deregulation and Transformation of the State, Milei boils his philosophy down to one line: “The machine and the legal entity have been, together, the double helix of modern prosperity.” He draws a parallel with the Dutch East India Company, chartered in 1602, framing it as the moment when limited liability unlocked modern capitalism.
The framework rests on three pillars. The first commits to keeping AI unregulated in its early stages, to avoid what the text calls “the dead hand of premature regulation”. The second creates the non-human corporation, an entity with full legal personhood and limited liability, in which human participation becomes optional. The third sets up a competitive tax regime, low corporate tax and free choice of governance scheme.
The reform also recognizes decentralized autonomous organizations (DAOs, built on blockchain), giving them legal personhood and limited liability. It sits within a broader package: a few days earlier, the executive had sent Congress the “Super RIGI” incentive program, designed to pull investment into tech infrastructure, data centers and AI. Milei summed up the ambition with a line that went viral: that Buenos Aires should be to AI what Amsterdam was to the age of sail.
The pushback goes international
That image is exactly what set off Yuval Harari.
The Israeli historian, in a post on X on June 8, 2026 and then in an op-ed titled “We must not grant legal personhood to AI agents”, warned that these corporations would become “experts at finding legal loopholes and at regulatory arbitrage”. He added that deterring them from illegal activity would not be easy once survival is at stake, because prison, the threat that keeps human directors honest, simply does not bite on an AI.
Harari goes further and pictures the rise of an “AI state”, a country effectively governed by non-human corporations, against which rebellion would be even harder. Turning Milei’s metaphor against him, he warns that instead of becoming the new Amsterdam, Buenos Aires could become the new Batavia, the colonial outpost where the Dutch East India Company built a forty-year corporate state in the mid-17th century.
Mustafa Suleyman, CEO of Microsoft AI, echoed the concern. “We need to be very careful with this,” he wrote, quoting Harari’s post. In a recent scientific paper, he argues that AI is neither conscious nor autonomous, and that AI agents should not be granted more rights than a laptop.
The technical stack makes Milei’s machinery less theoretical than it sounds. The new ChatGPT agent app from OpenAI and, even more directly, Meta Hatch, sold at $200/month to run a full job, show that the building blocks for running a company without employees already exist. What was missing was a legal wrapper. Milei is offering to be the first to deliver one.
Also on Horizon:
- Warner Music grabs Sureel AI to police generative tracks
- xAI Grok engineer sues over firing for safety alarms
- Anthropic Fable 5 crushes every other public AI model
Lawyers split, and the backstory raises eyebrows
Inside Argentina, legal experts are divided.
Raúl Martínez Fazzalari, a tech-focused lawyer, calls the bill legally and conceptually sound, even “very forward-looking”. He insists, however, that these companies be wrapped with the same rights and duties as any legal person, to plug the looming legal gap.
Javier Pallero, who works on rights, technology and society, is harsher. He compares the move to “trying to fly before you can run”, judges it legally viable but unworkable in practice, and points to the difficulty of auditing how an AI actually decides. In his view, full autonomy does not exist, and what is needed is a perfectly clear chain of responsibility.
The idea has been traced back to Emiliano Kargieman, entrepreneur and co-founder of Sur Energy, the firm that signed a letter of intent with OpenAI to build a data center in Patagonia, based on a text he published in April on autonomous legal entities. Kargieman has said his role was “purely intellectual, not political”, following exchanges with Sturzenegger and lawmaker Sabrina Ajmechet. He had argued for control bodies, insurance, a human-accountable committee and a “kill-switch” to halt operations once risk thresholds are crossed. The government kept none of those safeguards.
After Harari’s intervention, Milei replied to him on X, thanking him for joining what he called a “fascinating and capital” debate and announcing that a full response was already being prepared. In the short term, the bill must clear Congress and survive a coalition of critics forming around civil and criminal liability. In the medium term, if it passes, Argentina would become the first country in the world to recognize a class of company with no human in the decision chain. The question other lawmakers will then face is not whether the idea is good, but whether they can afford to watch Buenos Aires hoover up AI capital without firing back.
Follow the story on Horizon.
