A retreat, not a defeat
On August 19, 2026, ruling-coalition Senate leader Patricia Bullrich told the General Legislation Commission that Argentina's overhaul of corporate law would no longer permit companies run entirely by artificial intelligence with zero human oversight. The concession applies to sociedades automatizadas (automated companies) and DAOs created under Deregulation Minister Federico Sturzenegger's bill to replace Ley 19.550, the 1972 statute that has governed Argentine corporate structure for more than half a century (La Nación; Infobae). Bullrich's own words framed it as a legal-certainty fix, not an ideological reversal: "Pensamos que va a dar más seguridad jurídica que el órgano de administración esté integrado al menos por una persona jurídica con idoneidad" — a concrete, human party so that responsibility doesn't, in her phrase, "quede en la nada."
What the original bill actually proposed
The bill, filed as expediente PE-193/26 and first presented to the Senate on June 24, 2026 by Sturzenegger alongside Inspector General of Justice Alejandro Ramírez and Undersecretary Paula Taddei Farfán, would repeal Ley 19.550 entirely and replace it with a 277-article framework built around five pillars: contractual freedom, registry digitization, modernized company types, sharper administrator accountability, and — the headline-grabbing piece — formal recognition of Decentralized Autonomous Organizations operating on blockchain and companies making administrative decisions through algorithms (Argentina.gob.ar; Senado official record). Ramírez's justification was blunt: the current law was "concebida para una economía industrial previa a Internet" — built for an industrial economy that predates the internet, let alone generative AI.
Steelmanning the objection
The critics deserve a fair hearing before dismissal. Former IGJ director Ricardo Nissen, appointed under the previous Kirchnerist government, argued in Senate testimony that stripping human accountability from corporate structures amounts to "the consecration of commercial societies as a trap" — a vehicle for capital to escape liability while workers and creditors are left holding the risk (Perfil). That is not a frivolous concern. An AI system directing a company's assets, entering contracts, or restructuring debt with literally no natural or legal person answerable for the decision creates a genuine accountability vacuum — one that Argentina's own bankruptcy, tax, and anti-money-laundering regimes are not built to address. A company that can point to "the algorithm" when a contract goes bad, a debt goes unpaid, or sanctions screening fails is not innovation; it is a liability shield with better branding. Regulators who want a named, reachable, legally responsible party behind every corporate decision are not being reactionary — they are asking for the baseline that makes corporate law enforceable at all.
Why the compromise is still the right outcome
What's notable is that Bullrich's fix addresses exactly that gap without gutting the bill's actual innovation. The revised text keeps AI-driven administration, algorithmic decision-making, tokenized equity, and blockchain-native DAOs — it simply insists a human or legal person sit inside the administrative body as the accountable party. That is a proportionate fix, not a rollback. It mirrors how most functioning jurisdictions have approached algorithmic governance: Wyoming's 2021 DAO LLC statute and similar frameworks elsewhere require a registered agent precisely so that liability doesn't dissolve into code. Argentina's DAO provisions, per reporting on the bill's text, already required identifiable legal representatives and traceability compliance even before this concession — so the change mainly closes the loophole for sociedades automatizadas that aren't structured as DAOs (El Cronista).
The deregulatory core survives intact: full digital incorporation, elimination of mandatory paper records, a national companies registry, "electronic headquarters" for legal notice, and — critically — a shift from mandatory to suppletive rules, meaning company bylaws now prevail over statutory defaults unless the bylaws say otherwise. That is a real modernization of a 54-year-old code, not cosmetic tinkering.
The lesson for AI corporate law elsewhere
Argentina is one of the first jurisdictions to legislate corporate personhood questions for AI-run entities from first principles, rather than patching common-law fictions onto existing statutes. The Senate's amendment is instructive precisely because it didn't need to choose between AI-native corporate forms and accountability — it required both. Jurisdictions watching Buenos Aires (the EU, in particular, as it drafts AI Act implementing rules touching automated decision systems) should note that the fight here was never really about whether algorithms can run companies. It was about who answers for it when they do. Argentina's Senate got that balance closer to right than the original bill did — which is a rare case of legislative friction actually improving a text rather than just slowing it down.
The General Legislation Commission, chaired by Senator Nadia Márquez, has not yet released final consolidated text; more expert testimony was scheduled through late August 2026. The bill's fate — and the precise wording of the human-responsibility clause — remains to be finalized before a floor vote.