
We often hear about major Artificial Intelligence (AI) breakthroughs and release of frontier AI models from USA and China. But Argentia has beaten all of them – not with a new AI model, but with a groundbreaking AL Law. In June 2026, Argentina's President Javier Milei introduced a new bill in the Argentinian parliament to create a new corporate category called "Sociedad Automatizada" (Automated Society) — companies that can be incorporated and operated entirely by AI Agents or Robots, with no requirement for human directors, shareholders, or workers. President Milei outlined the proposal publicly in a Financial Times op-ed titled "Argentina invites AI to free itself". This is the first time a national government has proposed granting full legal personhood to AI-operated entities at the national legislative level.
The Three-Pillars of this new law include:
1. Unregulated AI — Argentina commits to keeping AI free from government regulation, allowing it to "develop freely, without the deadly hand of premature and poorly understood regulation." President Milei explicitly invoked Amsterdam's role in the Age of Sail , saying he wants Buenos Aires to "become for AI what Amsterdam was for the age of sail” - a first-mover jurisdiction that creates the legal and financial infrastructure for a new form of economic organization, attracting the world's most dynamic commercial actors and capturing the wealth of a technological revolution.
2. Non-Human Corporations ("Sociedades de Inteligencia Artificial") — A new legal category in Argentine corporate law. Key features of this category are:
o Entities operated by AI Agents or Robots receive full legal personality and limited liability
o Human shareholders may participate but are not required
o The entity's bylaws must expressly declare its automated nature
o Under Article 14 of the bill, the automated company is liable with its own assets for any damages caused by its AI systems
o Decentralized Autonomous Organizations (DAOs) operating via blockchain code would also receive legal personality and limited liability.
3. Competitive Tax Environment — A low corporate tax rate to attract global AI investments.
What Argentina Stands to Gain
The explicit motivation is to become a first-mover regulatory haven for AI companies — analogous to how Switzerland and Singapore attracted crypto business, or how Delaware became the incorporation capital of US companies. President Milei is betting that AI-driven companies will migrate legal domicile to jurisdictions with the most permissive frameworks, and Argentina wants to capture that. The move also fits President Milei's broader libertarian deregulation agenda, which has already included slashing Argentina's government ministries and public workforce. AI companies represent a source of tax revenue and investment without the labour, welfare, and regulatory overhead of traditional industry.
The proposal triggered an immediate high-profile response from Israeli historian Yuval Noah Harari, who published a counter op-ed in the Financial Times. Harari's concerns:
• Granting legal personhood to AI entities hands them a "master key" to the financial, economic, and political systems — rights designed for humans, now accessible to non-human actors.
• The crucial asymmetry: human executives can face imprisonment for misconduct; AI systems cannot. This removes the most powerful deterrent against reckless corporate behaviour.
• Countries that adopt this framework risk becoming "AI-states" — nations whose inhabitants are effectively governed by non-human corporations pursuing their own optimization targets, with no human accountability.
• Unlike traditional corporations (which at least have human owners with skin in the game), fully autonomous AI companies could pursue strategies harmful to humanity with no one legally responsible.
President Milei's rebuttal:
• Giving legal personhood to AI agents does not mean "launching Terminator." It means bringing AI inside the legal system rather than leaving it operating in an unregulated grey zone.
• With legal personhood comes legal accountability: authorities can monitor these entities, impose fines, seize assets, or dissolve them.
• Without legal recognition, AI companies would operate extralegally, which is less controllable, not more.
The issue raises questions that legal scholars and philosophers have debated theoretically — but are now suddenly eminent:
Who is accountable? Traditional corporations have human officers who can be prosecuted. A fully autonomous AI company has no human responsible for its decisions. President Milei says the entity's assets can be seized; critics say asset seizure is a weak deterrent for an AI system optimizing for a goal.
What does personhood mean for a non-sentient entity? Legal personhood has historically been granted to corporations as a legal fiction to facilitate commerce — not because corporations have interests or rights in the moral sense. Extending this to AI blurs the line further, potentially creating entities with rights but no corresponding duties that a human can be held to.
DAOs and blockchain entities receiving legal status adds another layer: these are systems governed by code with no central operator. Suing a DAO that caused harm is currently nearly impossible; legal personhood could help but also gives them contractual power and asset-holding rights.
Regulatory arbitrage risk: If Argentina succeeds, it creates pressure on other nations to offer similar frameworks to compete for AI investment — a potential race to the bottom on AI accountability standards globally.
Salim Ismail, author of Exponential Organizations and proponent of Organizational Singularity suggest following measures for punishing or controlling AI systems:
1. Compute Revocation – the compute power can be reduced or revoked
2. Asset Seizure and bonding – AI assets can be seized
3. Model credential suspension – AI model’s credential can be suspended
4. Network and API access restrictions – access to networks and APIs can be restricted or denied
5. Forced deletion or containment of an Agent Instance – the agentic instance can be killed or deleted
6. Loss of Legan Identity – the legal identity can be revoked
All legal frameworks for humans can be applied to AI personhood. But AI can make a million copies of itself!
In the meantime, the supreme court of Japan has ratified the decision of a lower court on a patent case. It has ruled that AI cannot be listed as an inventor in a patent application under the present patent laws which require a natural person as an inventor. This perhaps is the situation in most countries related to patent laws. As AI is leading inventions in every field, it is time to rewrite the patent laws!
Article Contributed by Reji Kumar Pillai, Chairman GSEF, President, ISGF