Argentina's AI-driven companies still rely on human input.

      The government of Argentina aims to allow a company to operate without a designated individual in charge and has proposed a bill to legalize this. The proposal introduces a new category known as the non-human corporation, which would be managed by AI agents or robots capable of signing contracts and holding assets independently, with human shareholders being optional.

      However, upon deeper examination, the bill appears to rely more on human oversight than suggested by President Javier Milei, especially at a time when many governments are tightening regulations on AI rather than loosening them. Even in its most autonomous model, based on a blockchain framework designed like a decentralized autonomous organization, the entity still necessitates a human legal representative for any actions requiring a personal signature, as well as a human promoter who is bound indefinitely for the company’s responsibilities at its inception, according to legal assessments of the proposal.

      Additionally, in areas where anti-money-laundering regulations are applicable, a human compliance officer is also mandatory. The government submitted this draft in May, aiming to replace Argentina’s general corporations law from 1972. Officials have positioned the reform on three foundations: keeping AI itself free from regulation, establishing the non-human corporation category, and implementing a low corporate tax rate to attract technology investments to Buenos Aires.

      This initiative is distinct from Super RIGI, the government’s complementary incentive program for large AI data centers, although both share a common target audience of investors. The proposal is competitive, as Milei’s administration seeks to establish Argentina as the preferred location for AI enterprises that would rather avoid regulatory scrutiny or the requirement of a human board. Whether this ambition holds up against the bill’s provisions on liability remains uncertain.

      Legal commentary on the draft has pointed out that even the supposedly autonomous structures maintain what one analysis described as a human floor, as a director configuring or overseeing an AI system remains accountable for its actions. The proposal has faced criticism from historian Yuval Noah Harari, who argued that eliminating a clearly accountable human from company decisions creates the kind of liability gap that corporate law intends to prevent. Harari referenced Milei’s comparison of the initiative to the Dutch East India Company, highlighting that one of its most significant actions was the destruction of the port of Jayakarta in 1619, leading to a private empire in the region. He cautioned that Buenos Aires risks becoming a “new Batavia” instead of a financial center.

      Microsoft AI CEO Mustafa Suleyman echoed this sentiment, asserting that AI agents should not have more legal standing than a computer and suggesting that developers should resist the illusion that their systems are quasi-persons entitled to rights. Milei offered a comprehensive response on social media, contending that establishing a defined legal category for AI-operated entities would facilitate regulation, as it would provide regulators with a named structure to address rather than an anonymous piece of software functioning outside existing laws.

      This argument hinges on whether the classifications will remain valid once AI-run entities scale up, a scenario that has not been tested in any jurisdiction. Critics argue that the direct threats that keep human executives accountable, primarily the risk of prosecution, are ineffective against algorithms, and a promoter’s unlimited liability may be insufficient when the entity begins making decisions that are not fully understood.

      It remains uncertain whether Argentina’s Congress will approve the bill as proposed or if other governments will adopt similar measures. The discussion comes as regulators in other regions are moving in the opposite direction, tightening rules regarding autonomous systems. Advocacy groups pushing for the EU’s AI regulations seek increased human oversight of automated decisions, while companies like SAP are restructuring executive oversight of AI instead of eliminating it. Argentina’s bill represents a gamble that this contrary approach will yield immediate benefits, although a human presence is still required behind it, regardless of the marketing narrative.