A Radical Bill Meets a Modest Correction
Argentina's government set out to rewrite corporate law for the algorithmic age. On June 24, 2026, Deregulation Minister Federico Sturzenegger presented the executive branch's proposal to replace Law No. 19,550 — the General Corporations Law that has governed Argentine business since 1972 — before the Senate's General Legislation Commission, chaired by Senator Nadia Márquez (senado.gob.ar). The centerpiece was a new legal category: the sociedad automatizada, a company that could operate "completely autonomously through an algorithm or AI, without human intervention in daily operations," carrying full legal personality and limited liability (Infobae).
By August 19, 2026, that proposal had been walked back. Senator Patricia Bullrich announced the ruling coalition would amend the bill so that automated companies and decentralized autonomous organizations (DAOs) must have "at least one physical or legal person responsible," with the administrative body required to include someone with "sufficient competency for control of automated systems" (El Argentino Diario). The zero-human-oversight version of the bill is dead. A human-supervised version is what will move forward.
The Case Critics Made — and Why It Was Right
Before arguing against overcorrection, it's worth taking the objection seriously, because the objection was correct. Digital-rights analyst Javier Pallero laid out the sharpest version of it: if an automated company can commit fraud, breach contracts, or cause harm, and "responsibility remains limited only to the entity's assets" with no natural person answering criminally, the corporate form becomes a liability shield with nobody behind it (Infobae). That is not a hypothetical concern about future AI risk — it is a concrete gap in how tort, contract, and criminal law attach to conduct. Courts, creditors, and regulators need someone to serve process on, someone whose conduct can be judged negligent or reckless, someone who can be deposed. An algorithm cannot testify.
Pallero's second point — the "black box problem" — also holds up: if algorithmic decision-making is opaque enough that even the people who built it cannot fully reconstruct why it acted as it did, then a company with no human overseer isn't just under-regulated, it's unauditable by design. Historian Yuval Noah Harari pushed a related, more sweeping version of the argument in his public exchange with President Javier Milei, warning that granting AI systems legal personality without a human anchor hands them "a master key" to financial and economic systems (La Nación). Milei's response — that legal personality is simply "a container of responsibility," no different in kind from the corporate structures that enabled capitalist expansion since the Dutch East India Company — is a fair analogy, but it elides the actual mechanism of how responsibility gets assigned when the entity making decisions isn't a person.
Why the Fix, Not the Original Bill, Is the Real Innovation
That said, the broader reform remains worth defending, and the Senate's amendment is precisely the kind of proportionate correction that lets innovation proceed without the accountability vacuum. The government's own framing is correct on the merits: a corporate statute drafted in 1972, before the internet, before platform economies, and decades before generative AI, genuinely constrains how Argentine businesses can organize themselves around software-driven operations (Argentina.gob.ar). Digital incorporation, a national companies registry, and a legal wrapper for DAOs are unambiguous modernizations that most jurisdictions will eventually need. Sturzenegger's own numbers illustrate the stakes: he has projected that up to 50 million autonomous software agents could be operating out of Argentina within a decade, drawn by a permissive but clear legal regime (Infobae). That's a genuine first-mover opportunity — but only if the regime is durable, and a regime built on an accountability gap invites exactly the kind of fraud scandal or judicial reversal that would kill the category outright within its first few years.
The Senate's fix threads that needle. It does not re-impose the old law's assumption that every company needs a full human board or a minimum headcount — the reform's other four pillars (autonomy of will, digitalization, modernized corporate types, technological updating) survive intact, according to the government's own five-pillar summary of the bill (Senado de la Nación). It adds exactly one constraint: a named person who is accountable when the algorithm is not. That is not the heavy-handed, capability-based prohibition some legislators elsewhere are now proposing for advanced AI systems generally — Senator Bernie Sanders and Representative Greg Casar's newly introduced Ban Artificial Superintelligence Act in the US Congress would attempt to define and outlaw whole classes of AI capability outright (MediaNama). Argentina's approach is narrower and more administrable: it doesn't try to define which AI systems are too dangerous to exist, it simply insists that whoever deploys one in corporate form must name a person who owns the consequences.
The Precedent That Matters
What Argentina has actually produced, whether or not the automated-company category ever attracts meaningful capital, is a template other legislatures drafting AI corporate rules will study: don't ban the structure, anchor it. A named human or legal person with "sufficient competency" to supervise an automated system is a testable, enforceable standard — regulators can ask whether that person actually exercised oversight, plaintiffs can sue them, prosecutors can charge them. That is a far more workable standard than either extreme: neither Sturzenegger's original zero-oversight draft nor a blanket capability ban of the kind now advancing in Washington. The Senate commission continues its review in subsequent sessions; the amended text of the human-oversight requirement has not yet been formally voted into the bill, and the DAO liability provisions still need to be reconciled with existing civil and criminal codes. But the direction — proportionate, accountability-anchored regulation rather than prohibition — is the right one, and Argentina got there through legislative friction rather than executive fiat.