On 4 October 2026, Argentina became one of 17 endorsers of the Kyoto Vision for a Golden Age of Science. The White House Office of Science and Technology Policy, led by Director Michael Kratsios, launched the declaration at the 23rd STS Forum in Kyoto. Buenos Aires signed on while the Milei government is courting AI investment with a promise not to regulate the technology preventively. Taken together, the two moves are a coherent strategy, but the declaration only covers the easier half of it.
What the declaration actually commits to
The White House release lists the endorsers as Argentina, Bulgaria, Chile, Cyprus, Germany, Greece, Indonesia, Italy, Japan, Kazakhstan, South Korea, New Zealand, Poland, Singapore, the UAE, the United Kingdom and the United States. The Kyoto Vision text says signatories aim to expand access to "SI tools, scientific data, computing infrastructure" and experimental facilities. Here "SI" means "super intelligence", the document's own term. It also calls for funding systems that reward ambitious ideas, through "long-duration awards, rapid grants, and prizes and challenges", and for "new institutional models for science". A section is headed "Harnessing Super Intelligence for Science".
The text is a statement of shared direction, not a treaty. Nothing in the material we reviewed creates a binding obligation, a funding schedule or a compliance mechanism. That is typical of such declarations. The value for Argentina is mostly signalling and network access: alignment with Washington's science agenda, and a seat in the conversation on shared research compute and data.
The strongest case for regulating first
Critics of Argentina's approach have a serious argument. Without rules, they say, a country with cheap energy and weak bargaining power becomes a host for other people's infrastructure, bearing the costs while the benefits go abroad. Nicolás Wolovick of the Universidad Nacional de Córdoba told AP that without rules Argentina would become "un laboratorio", according to Euronews's Spanish edition. The same report says the proposed incentives do not require compute to be reserved for local use, nor reporting of energy and water consumption. Residents of Trelew have started a campaign called "Trelew libre de centros de datos".
The scale makes the concern concrete. AP reports that Polish renewables firm Green Capital plans a US$3 billion data centre in Patagonia, with 14 landowners leasing about 130,000 hectares. At full demand the campus would use roughly 2.7 times the electricity Buenos Aires city used in 2025, by AP's calculation. The project has no client yet, and construction is not expected before about 2028.
Where the deregulatory case is stronger
Still, the case against a comprehensive AI statute is strong on its own terms. Milei told an audience in New York, per Bloomberg Línea, "En Argentina no le vamos a poner un cepo a la Inteligencia artificial." Horizontal AI laws written before the technology settles tend to lock in definitions that age badly. They also impose compliance costs that fall hardest on small firms and local developers, who cannot absorb them the way a multinational can. A country that already regulates through consumer protection, data protection, competition and civil liability can let those tools handle most harms and reserve new rules for demonstrated gaps.
The Kyoto Vision fits that logic. It asks governments to remove "unnecessary administrative burdens" for scientists and widen access to tools and data. Those are enabling measures, and they cost little. A researcher at a public university who can reach shared compute and datasets benefits regardless of how data-centre tax policy resolves.
The gap between the signature and the plan
The weakness is that "no preventive regulation" is not the same as "no governing framework". Julia Pomares, a regional director at GIDE and a member of a UNESCO expert body on AI, told Bloomberg Línea that Argentina's personal data protection law, enacted in 2000, should be updated first. She also flagged the executive's company-law reform, which may create "sociedades autónomas" (autonomous companies). That raises the question of who is responsible when an AI system acts, and she argued final responsibility must stay with a human.
This is the proportionate-regulation point. A light-touch regime is credible to investors only if basic rules are predictable: who owns data, who is liable for harm, and who can enforce. Open scientific data, which the Kyoto Vision promotes, also depends on trusted privacy rules. Researchers cannot share health or administrative records at scale under a 26-year-old statute that predates machine learning.
The incentive regime is untested
The tax proposal is also unfinished. Euronews, citing AP, reports that the corporate income tax rate would fall from 25% to 15% for AI and other emerging-industry projects of at least US$1 billion. The measure is still pending in the Senate. The same report says opposition in Congress forced Milei to drop a bill easing foreign land-ownership limits and stalled his plan to let AI agents control companies. Congress has shown it will push back on the government's more aggressive proposals.
A sounder sequence would keep the investment pitch and add three low-burden conditions. First, require disclosure of energy and water use for large facilities, so local communities can judge the trade-offs. Second, reserve a modest share of compute for national research institutions, which would give the Kyoto commitments on researcher access real content. Third, modernise the 2000 data protection law before the new regime scales.
None of this requires a European-style AI Act. It requires the boring institutional work that makes deregulation durable. Argentina's signature in Kyoto is a reasonable bet on open science. Whether it becomes more than a signature depends on what Congress and the executive build at home.