Analysis-Argentina’s plan for AI-run companies can’t avoid humans

By Leila Miller and Kenrick Cai

BUENOS AIRES, July 3 (Reuters) – Argentine President Javier Milei generated both excitement and fear last month when he announced a congressional bill to create “non-human corporations” run by AI, but the companies would actually require human involvement.

Milei described in a Financial Times op-ed piece a new type of company that could run without human employees, in which AI agents or robots would “exercise independent judgment in unpredictable environments.” Argentina would become the first country to pass legislation creating a category for companies run by AI, several legal experts said.

“We are open for business,” Milei declared, sparking criticism from Israeli historian Yuval Noah Harari, who warned giving AI too much power may reduce corporate accountability.

But the reality is less revolutionary, corporate attorneys said.  The “automated company” introduced in the proposed reform, part of a comprehensive bill seeking to modernize and cut bureaucracy in corporate law, would be required to have a human administrator to oversee operations. 

The bill also allows a company’s administration to use AI for decision-making without exempting the administrators from supervising the outcome.

It would be “too wild a first step to dispense with human agency entirely,” said Lawrence Cunningham, director of the Weinberg Center for Corporate Governance at the University of Delaware. Still, Cunningham called Milei’s proposal bold.

“We’re not changing the world here so much as we’re recognizing that you might run a business without any HR,” he said. “It’s the beginning of something.”

Diego Duprat, a law professor and co-author of the bill, said automated companies already exist in some forms, referring to AI-supported cashier-less supermarkets.

The company would be liable for damages caused by AI or algorithmic systems, the bill states.

A representative for the office of the presidential spokesperson said that there are currently no companies or commitments of investment that are connected to the bill.

“What is happening is that we are proposing something innovative, aimed at making Argentina an attractive jurisdiction for the establishment of automated companies,” the representative said in a statement. “This project is key to creating better conditions for attracting investment.”

Milei, who has brought inflation down sharply and sought to attract foreign investors with incentives, has repeatedly pitched Argentina as a future AI hub, highlighting Patagonia’s cold weather and energy supply as ideal for data centers. OpenAI and Sur Energy announced plans in October for a data center with an investment of up to $25 billion.

‘PREDICTABLE FRAMEWORK’

Merely having a law that refers to a company’s central use of AI may attract investors, said Maria Gisele Cano, a corporate attorney in the province of Buenos Aires. She has received more than a dozen inquiries from entrepreneurs in Argentina and abroad about the proposal.

“These companies will have a clearer and more predictable framework for conducting their operations in this environment,” she said.

Yonathan Arbel, a professor who researches AI at the University of Alabama’s law school, said Argentina could gain a “huge competitive advantage” if it creates a welcoming environment for AI business. He said the bill could benefit from specifying that AI agents should have a digital ID for interactions with people and companies.

The proposal also allows for the creation of companies that are decentralized autonomous organizations (DAOs), built on blockchain, enabling members to vote on proposals with digital tokens.

Argentina is a top Latin American cryptocurrency market. Ricardo Mihura Estrada, former president of Bitcoin Argentina, said the proposal’s requirement that token users be identified and registered would be a challenge for an industry built around anonymity.

“I think it’s well intentioned, but I see difficulty in it being adopted in the blockchain world,” he said.

The representative for the presidential spokesperson’s office said that identifying token users is a minimum security requirement, adding: “DAOs that prefer to maintain a completely anonymous structure may continue to operate outside this regime, but they will not gain access to the legal benefits it offers.”

VIEW FROM SILICON VALLEY

Milei’s automated companies mirror a vision of OpenAI CEO Sam Altman, who said in 2024 that AI will enable a company with a single employee to reach a $1 billion valuation.

Several U.S. states, including Texas and Utah, have set up legal frameworks for businesses to experiment with AI, said Emerald Greywoode, a researcher at the Weinberg Center. These may include guidelines that an AI business receives more human oversight at the start of such testing.

Current technical capabilities are not advanced enough for AI agents to make fully autonomous business decisions, experts say. But Silicon Valley entrepreneurs are increasingly shifting their budgets from hiring employees to spending on AI computing power to perform the same tasks, according to Basis Set Ventures managing partner Lan Xuezhao, who invests in AI startups.

AI entrepreneurs are most concerned about access to and costs of computing power, chips, and energy, she said, adding that light-touch regulation could become attractive as regulatory bodies in the United States and Europe impose stricter rules around AI use.

Still, Milei’s bill alone is unlikely to turn Argentina into an AI hub, Lan said.

“The most important thing is if the talent goes to Argentina,” she said. “People will follow.”

(Reporting by Leila Miller from Buenos Aires and Kenrick Cai from San Francisco; Editing by Rod Nickel)