Imagine a company with no employees. Software handles customer service, negotiates with suppliers, issues invoices and pays taxes, and nobody is on payroll. That’s what Javier Milei’s government is proposing with the “Automated Company,” a new legal entity that has been sitting in the Senate since June and is now stalled.
The idea was unveiled on April 28 by Federico Sturzenegger, the Minister of Deregulation and State Transformation, at Expo EFI, an economics and finance conference. His argument: if a large share of the world economy comes to be produced by AI agents—programs capable of carrying out tasks on their own—then it makes sense for those companies to register here. “We could have 50 million people and 500 million AI agents incorporated here, producing for the world and paying taxes in our country,” he said, according to Ámbito.
What the bill says
A month later, on May 29, the Executive Branch sent Congress a General Companies Law bill that would replace the current law, dating back to 1972. It runs more than 270 articles, and just one of them, Article 14, would create the Automated Company: a company of any type that operates “through autonomous algorithmic systems or artificial intelligence agents, without requiring employees in a formal employment relationship.” Its bylaws would have to state this, and its name would have to include the word “Automated.”
The obvious question is who’s liable if one of those agents defrauds a customer or breaches a contract. The bill says the company is liable “with its assets,” like any corporation. And even though the operation is automated, management still rests with people: Article 88 requires that administration be in the hands of natural or legal persons, and Article 102 clarifies that using AI to carry out operational functions or make decisions “does not exclude or limit the liability of administrators.”
Why it’s stalled
In the Senate, that explanation didn’t fly. On June 24, Sturzenegger defended the bill before the General Legislation Committee, where it has sat without a recommendation—the step required before a vote—since June 11. “There’s no reason to think it will be less safe,” he said, according to La Nación.
The government’s allies wanted more. On August 19, in order to secure their support, Patricia Bullrich, head of the La Libertad Avanza bloc in the Senate, announced that each automated company would have “physical persons responsible,” according to Infobae, or a legal entity “with sufficient competence” to oversee the systems, according to La Nación. That revised text still hasn’t been published. In September, more than 30 civil society organizations asked for the article to be removed. On September 29 and again on October 4, Infobae described the law as “on standby,” and it wasn’t on the agenda for the Senate’s October 29 session either. On October 20, at SAIAConf 2026, Sturzenegger presented the change as a settled compromise: “Supervision will be human, and operation, autonomous,” according to La Nación.
The same debate in the United States
Delaware, the small state where roughly two-thirds of the Fortune 500 are incorporated, has been working on a similar entity since 2025. Its draft came to light in June, after the Argentine bill, and wouldn’t be introduced in its legislature until 2027. It creates the AIC (artificial intelligence company), a business that can be directly managed by AI agents and whose owner, according to Ars Technica, “is not liable for any act or omission” of the company, except in cases like fraud.
That protection for owners set off alarms in Washington. On October 26, Republican congressman James Comer, who chairs the House Committee on Oversight and Government Reform, asked Delaware’s governor for answers, warning that the entity could shield AI companies “from liability for reckless harmful actions,” according to the committee’s statement. Delaware has until November 9 to respond.
Why it matters
Both bills promise the same thing: agents that sign contracts, sell and pay bills. For that to work, a bank or supplier needs to know who to go after before doing business with a company that has no employees. Our read is that the identifiable, responsible party the government’s Senate allies demanded could make this entity more trustworthy for anyone who has to do business with these companies.
What happens next will be decided in November. Delaware answers Comer on the 9th, and Argentina’s Congress closes its regular session on the 30th. If the revised text, with a committee recommendation, hasn’t surfaced by then, there will be no Automated Company this year.
Frequently asked questions
What is an Automated Company in Argentina?
The Automated Company is the legal entity that would be created by Article 14 of the General Companies Law bill, signed by Argentina’s Executive Branch on May 29, 2026 and submitted to the Senate on June 1: a company that operates using artificial intelligence agents, with no employees in a formal employment relationship.
Has Argentina already approved AI-run companies?
As of October 29, 2026, Argentina has not approved AI-run companies: the bill has been sitting in the Senate’s General Legislation Committee since June 11, without a recommendation. On August 19, Patricia Bullrich announced that the ruling party would require identifiable parties responsible—either natural persons or a legal entity “with sufficient competence”—but that text was never published. It also wasn’t on the agenda for the October 29 session. Regular sessions end on November 30.
Who is liable if an AI agent at an Automated Company causes harm?
Under Argentina’s bill, the Automated Company is liable for damages caused by its AI agents with its own assets, just like a corporation. Administrators, who must be natural or legal persons, remain liable even if they use AI to make decisions. To secure its allies’ support, the ruling party promised to add an identifiable party responsible, a change Patricia Bullrich announced in August 2026.
Are there similar laws in other countries?
Delaware, the U.S. state where roughly two-thirds of the Fortune 500 are incorporated, is preparing the AIC (artificial intelligence company), a business that AI agents could manage. Its draft came to light in June 2026, after the Argentine bill, and wouldn’t be introduced before 2027. Congressman James Comer opened an investigation into the proposal on October 26.