When President Javier Milei and Deregulation Minister Federico Sturzenegger took to the Financial Times on June 4, 2026, their pitch was unusually direct for a government op-ed: Argentina would keep artificial intelligence "unregulated," create a new legal category allowing AI agents to run corporations without human directors, and offer the most competitive fiscal environment on Earth for tech investment. "We are open for business," they wrote, invoking the Dutch East India Company as a historical parallel and pledging to make Buenos Aires what Amsterdam was "for the age of sail navigation."
The response was swift and unusually high-profile. Historian Yuval Noah Harari published a counter-essay in the FT on June 9, warning that countries granting legal personhood to AI risk becoming "AI-states, countries whose inhabitants could be governed by non-human corporations." Mustafa Suleyman, CEO of Microsoft AI, endorsed Harari's position publicly, writing "we have to be very careful about this" and citing his own Nature paper arguing AI agents should hold "no more rights or freedoms than a laptop computer." For a single op-ed from a mid-sized South American economy to trigger responses from a philosopher-historian and a major AI executive within days is, itself, evidence that Argentina touched a nerve.
What the Three Pillars Actually Mean
The strategic framework has three components, each with distinct implications.
The first — keeping AI development free from "the deadly hand of premature and poorly understood regulation" — is less a policy than a commitment to abstain from one. Argentina will not enact broad sector-wide AI acts of the kind advanced in Brussels. For a country with stretched regulatory institutions, this has a pragmatic logic: poorly designed rules, badly enforced, often produce worse outcomes than none at all.
The second pillar is structural. Before the op-ed, the government sent Congress a bill amending Law 19.550 — Argentina's corporate law since 1972 — to create the sociedad automatizada. These are entities "operating exclusively through autonomous algorithmic systems or AI agents," able to take the form of a standard S.A., S.R.L., or the newer S.A.S. They may have human shareholders, but human directors are not required. Corporate assets back liabilities; AI agents themselves bear no criminal exposure. The bill also explicitly recognises DAOs operating on blockchain networks. Companies must declare their automated status in their bylaws and append "Automatizada" to their registered name — a transparency measure, albeit a thin one.
The third pillar is fiscal: a "Super RIGI" extension of Argentina's existing investment incentive regime, offering stable, favorable tax treatment for large technology and infrastructure projects over an extended horizon.
The Investment Case Is Real
Before dismissing this as ideological performance, consider what it is already attracting. OpenAI and energy firm Sur Energy signed a Letter of Intent — announced at Casa Rosada — for a data centre in Patagonia capable of up to 500 MW and a potential investment of up to $25 billion, structured under Argentina's RIGI framework. That would rank among the largest technology infrastructure commitments in the country's history. OpenAI's chief global affairs officer flew to Buenos Aires to sign it.
Argentina is also deploying AI inside its own state. MIA — the country's first government AI agent, built on Meta's open-source Llama 4 — launched to serve the 26 million users of the Mi Argentina government application and handle more than 600 administrative procedures across 72 national agencies. Whatever one thinks of the deregulatory framing, this is a government that is using the technology it is championing, not just licensing it to investors.
Minister Sturzenegger's prediction that "in ten years, GDP will be made up of AI agents" is hyperbolic. But the underlying logic — that early-mover advantage in AI governance frameworks can attract capital the way incorporation-friendly laws attracted companies to Delaware — is not unreasonable for a country historically crowded out of high-value global investment flows.
The Governance Critique Deserves a Serious Answer
Harari's argument is not technophobia. It is a structural observation about enforcement. Human executives can be imprisoned; AI entities cannot. A sociedad automatizada whose assets are exhausted by one lawsuit cannot be compelled to behave differently in the future. If an AI-operated firm causes large-scale financial harm, systematically discriminates against users, or manipulates markets, the only recourse against the entity itself is civil litigation against its asset base. The bill as described provides no mechanism for suspension, mandated audit, or corrective operational requirements beyond what ordinary corporate law already offers any S.A.
This is not an argument against the category itself. It is an argument for what the category is currently missing: a liability floor, a mandatory incident-reporting obligation, and transparency requirements proportionate to the scale of risk the entity poses. These are not the regulatory burdens that stifle innovation. They are the conditions under which serious institutional capital — pension funds, sovereign wealth, major corporates — commits at scale rather than at Letter of Intent stage.
Suleyman's pushback, notably, did not come from a regulator or an NGO. It came from the CEO of one of the world's largest AI developers. His view — that AI systems should hold no more autonomy than a tool — reflects a broad industry consensus that anthropomorphising AI agents in law creates more problems than it solves, including for the companies building the agents.
What Comes Next
Peter Thiel's visits to Buenos Aires in April and May 2026, before the bill's Congressional submission, drew opposition accusations of regulatory capture. The Milei government denied any connection, and the bill's broad language does not read as a favour to any single firm. But the optics are a recurring vulnerability for an administration that frames its liberalisation as principled rather than transactional.
The sociedad automatizada bill is still moving through Congress. Argentina has a history of bold legal innovation that stalls in implementation — its fintech sandbox, its crypto framework, its open banking pilots all generated headlines before institutional friction slowed adoption.
What is not uncertain is that the question Argentina has forced onto the global agenda — how should law treat entities that act with commercial autonomy but cannot be held personally accountable? — will demand an answer in every jurisdiction. The EU will answer it one way, the US another, and Argentina is now offering a third. The debate Harari and Suleyman entered is not about Argentina specifically. It is about what comes next, everywhere.
Argentina's bet is that being first, and being permissive, wins the capital race. The smarter bet — consistent with pro-innovation, proportionate policy — is that being first with a credible liability framework wins more durably. The sociedad automatizada concept is sound. The accountability vacuum it currently contains is the one problem the Congress still has time to fix.