Argentina’s AI-run companies plan still can’t do without humans

Argentina’s government wants to let a company trade with no person in charge of it, and it has sent Congress a bill to make that legal.

The proposal creates a new category called the non-human corporation, an entity run by AI agents or robots that can sign contracts and hold assets on its own, with human shareholders optional rather than required.

But read past the headline framing, and the bill turns out to lean on humans more than President Javier Milei’s pitch suggests, at a moment when most governments are moving toward tighter AI governance rather than looser rules.

Its most autonomous format, a blockchain-recorded structure modelled on a decentralised autonomous organisation, still requires a human legal representative to bind the entity to any act that needs a person’s signature, plus a human promoter who answers without limit for the company’s obligations at formation, according to legal analyses of the draft.

Wherever anti-money-laundering rules apply, a human compliance officer is required too.The government submitted the draft in May, proposing to replace Argentina’s general corporations law, in force since 1972. Officials have pitched the reform as resting on three pillars: keeping AI itself free of regulation, creating the non-human corporation category, and a low corporate tax rate to draw technology investment to Buenos Aires.

It is a separate initiative from Super RIGI, the government’s parallel incentive package for large AI data centres, though both target the same investors.

The pitch is competitive. Milei’s government wants Argentina to become the jurisdiction of choice for AI ventures that would rather not answer to a regulator or maintain a human board. If that ambition survives contact with the bill’s own liability provisions is a separate question.

Legal commentary on the draft has noted that even its supposedly autonomous structures keep what one analysis called a human floor, since a director who configures or supervises an AI system remains answerable for what it does.

The proposal has drawn a pointed rebuttal from historian Yuval Noah Harari, who argued that removing a clearly accountable human from a company’s decisions creates exactly the kind of liability gap that corporate law exists to prevent.

Harari invoked Milei’s own comparison of the plan to the Dutch East India Company, noting that the company’s most consequential act was burning down the port of Jayakarta in 1619 and ruling the region as a private empire afterwards. He warned that Buenos Aires risked becoming a “new Batavia” rather than a financial hub.

Microsoft AI chief executive Mustafa Suleyman weighed in on the same side, writing that AI agents deserve no more legal standing than a laptop and citing his own recent essay arguing that developers should actively resist any illusion that their systems are quasi-persons deserving rights.

Milei responded at length on social media, arguing that giving AI-run entities a defined legal category would make them easier to regulate, not harder, since regulators would have a named structure to point at rather than an unaccountable piece of software operating in the shadows of existing law.

That argument depends on the categories holding up once AI-run entities operate at scale, something no jurisdiction has tested.

Critics counter that the deterrents keeping human executives in line, chiefly the threat of prosecution, mean little to an algorithm, and a promoter’s unlimited liability may prove thin once the entity is making decisions no person fully understands.

Whether Argentina’s Congress passes the bill as drafted, and whether other governments follow suit, remains open. The debate arrives as regulators elsewhere move the opposite way, tightening rather than loosening the rules around autonomous systems.

Advocacy groups pushing the EU’s AI rules want more human oversight of automated decisions, not less, and companies such as SAP have been restructuring executive oversight of AI rather than removing it.

Argentina’s bill is a wager that the opposite approach pays off first, with a person still standing behind it, whichever way the marketing reads.