On June 4, 1602, the Dutch East India Company introduced the limited liability company to the world. I guess there was no impact assessment or years of stakeholder consultations. Four centuries later, Javier Milei published an op-ed in the Financial Times with what I believe is the most ambitious pro-innovation legal proposal any government has made in years. Argentina plans to create a dedicated legal framework for AI development, including a new corporate category: the non-human corporation. This entity, operated by AI agents or robots, has the legal standing to sign contracts, hold assets, and function in the real economy. Buenos Aires, President Milei concluded, should become for artificial intelligence what Amsterdam was for the age of sail.
Predictably, this has already prompted worried commentary. Some academics and civil society groups have flagged the usual concerns: data protection, liability, and transparency. Yuval Harari weighed in from his customary perch above the clouds, warning that giving AI entities access to banks and contracts could be dangerous. One Buenos Aires Herald piece described the reaction among some Argentine experts as “concern.”
These are fair points. But consider the alternative framing offered by entrepreneur Martín Varsavsky: the law always runs ten years behind technology. A company is already a legal fiction. It does not breathe, vote, or have a soul. It exists to sign contracts, pay taxes, assume responsibilities, and be sued. If an AI-run company is already buying, selling, investing, and providing services, it is better that it has an address, capital, human accountability, and clear rules. The worst outcome is bots operating from personal accounts with nobody to hold responsible.
This is the logic behind what Milei and his deregulation minister Federico Sturzenegger are proposing. Three pillars:
- keep AI unregulated enough to develop freely without premature and poorly understood restrictions
- create the legal architecture for AI agents to operate
- establish a competitive fiscal environment for the companies that will run them.
At its core, it is a framework for accountability, the kind that comes from having a legal address rather than operating in the shadows.
Those of us who have spent years advocating for innovation-friendly policy have seen this film too many times. A technology emerges. Regulators panic. A framework is rushed through that was designed for technology from five years ago. The innovation migrates. Europe made this mistake with data (GDPR), is arguably making it again with AI (the EU AI Act), and will eventually import technology built elsewhere. As Adam Thierer noted, trying to preemptively plan for every worst-case scenario means many best-case scenarios never happen.
Argentina’s bet is different. Instead of banning by reflex, it is giving legal shape to a reality already arriving. The Dutch did not fully understand the joint-stock company either. The limited liability concept faced criticism for allowing wealthy men to gamble with other people’s money. They were not entirely wrong. But history showed the alternative was worse.
There is something the international commentary has mostly missed. This proposal comes from a country that spent two years dismantling one of the most suffocating regulatory environments in the Western hemisphere. Argentina advanced 20 positions in the Heritage Foundation’s Economic Freedom Index in 2024 and 2025, the largest improvement of any nation in both years. Fiscal surplus. Inflation controlled. A deregulation drive of genuine breadth. The AI framework is the logical extension of a coherent reform program.
Will it work? That depends on execution. The fiscal incentives need to be real, the legislation well-crafted, and Buenos Aires must deliver what it promises. On that last point, Argentina has a long history of reasons for doubt. But the reform program so far suggests something different is happening. The deregulation drive is genuine. The missing piece for a truly pro-innovation tax environment may already be on the table: a policy paper by Horacio Arana, Marianela Mendoza, and this writer proposes replacing Argentina’s labyrinth of more than 165 overlapping taxes with a single flat rate of 15 percent for individuals and businesses. Georgia quadrupled its tax revenue after adopting a similar model. Slovakia and Estonia saw comparable results. Companies deciding where to deploy AI infrastructure do not want to hire an army of accountants to navigate a tax maze. They want clarity. A flat tax could be the foundation that makes the AI framework credible. If Argentina delivers both, the offer to the world becomes compelling.
Nevertheless, skepticism about execution is different from dismissing the idea. The idea is good. In fact, it may be the most interesting governance experiment in AI policy anywhere in the world right now.
* Federico N. Fernández is a visionary leader dedicated to driving innovation and change. As the CEO of We Are Innovation, a global network of over 50 think tanks and NGOs, Federico champions innovative solutions worldwide. His expertise and passion for innovation have earned him recognition from prestigious publications such as The Economist, El País, Folha de São Paulo, and Newsweek. Federico has also delivered inspiring speeches and lectures across four continents, authored numerous scholarly articles, and co-edited several books on economics.
Source: We Are Innovation
