Milei Proposes Limited Liability for AI Agents and Tax Incentives in UN Speech
At the UN General Assembly on 23 September 2026, Argentine president Javier Milei unveiled a deliberately deregulated AI strategy: Argentina, he said, should stand at the forefront of AI freedom, and he pledged not to regulate the technology preemptively.
What Milei Said from the Podium
In his speech to the UN General Assembly in New York on 23 September 2026, Milei devoted significant attention to artificial intelligence. According to UPI, he said: "On artificial intelligence, Argentina will position itself at the forefront of freedom." The New York Times renders the statement in slightly different form — Argentina will stand at the forefront of freedom — and adds that Milei simultaneously asserted the country's "commitment to not regulating AI preemptively."
According to UPI, Milei rejected preemptive regulation of the technology and said Argentina would seek to position itself as a competitive location for AI development. He argued that the country has what it takes to participate in the AI value chain: critical minerals, energy, and skilled labor.
The Concrete Measures
This was more than rhetoric. According to UPI, Milei proposed a framework with two main elements:
- Limited liability for companies operating AI agents. This would reduce the legal risk of letting autonomous systems act on a company's behalf — one of the issues Western regulators treat most cautiously.
- Tax incentives for players that establish themselves in Argentina.
Beyond this, Milei has according to the NYT proposed creating "non-human companies" run by AI agents — an extension of the idea that liability and corporate form should be adapted to the technology, not the other way around. He is also working to lure foreign investment to data centers in Patagonia.
All of these points must be read as stated positions. No formally introduced bill or enacted Argentine policy is documented in the source material, and no company has confirmed commitments to build Patagonia data centers.
The Broader Program: AI in Government
Milei's AI enthusiasm does not stop at industrial policy. The NYT writes that he "vigorously posts AI-generated content on social media," and describes an AI-generated influencer close to the government known as "Libertarian Lucía" who was considered for a larger role in the government's communications.
When Milei's media strategist Santiago Oría was confronted with the fact that the influencer is a fabrication, he dismissed the concern: "It's totally inoffensive. There is nothing harmful," he said, according to the NYT.
The report must be read with caveats: the NYT article of 29 September 2026 is only partially available as a preview, so the depth of the paper's coverage of investors and the influencer episode cannot be verified beyond the excerpts. But the main point stands: Milei treats AI not only as an industry he wants to attract, but as a tool he himself uses in politics.
The Countercurrent at the Same Assembly
The contrast with the broader Western direction was striking precisely because it played out at the same assembly. According to Yahoo News' live coverage (The Telegraph), Burnham announced an AI defense partnership with the United States and said the world had to "go further" to ensure the safe development of the technology. The NYT describes the backdrop: while many Western leaders struggle to get control of a technology many fear could pose existential threats to humanity — and even tech executives are calling for safeguards — Milei is presenting Argentina as a haven without such rules.
Milei also tied the AI question into his broader critique of the UN. According to JURIST, he warned against potential UN regulation of artificial intelligence and called the UN a "bureaucratic leviathan," citing the UN's response to covid-19 as an example. The message was consistent: neither international bodies nor domestic regulators should set binding limits on AI before harm is documented.
Open Questions
There is considerable uncertainty about what Milei's initiatives will actually lead to:
- No enacted policy. The framework of limited liability, tax incentives, and non-human companies is, for now, only the president's stated intentions, not legislation. Whether such a framework would win a majority in Argentina's congress is unclear.
- No confirmed investors. No company has documented a commitment to data centers in Patagonia. The pitch is an offer, not a deal.
- Legally uncharted territory. Even if Argentina implements the proposal, the question remains how a liability system for AI agents would work in practice — and who bears the risk when limited liability meets autonomous systems.
What is nevertheless clear is that with this, Argentina has positioned itself as the most prominent Western example of AI deregulation as an active business strategy — in a way that stands out sharply from the mainstream in the same hall.