AI in Government: Where Mexico Stands on the World Map (2026)

Artificial intelligence is already operating inside the Mexican state. Not as a pilot experiment or a medium-term project: at least 14 federal agencies use AI systems in areas of auditing, security, and citizen services, according to data obtained through freedom-of-information requests. The problem is not whether the government uses AI, but that adoption is advancing at a different pace than the construction of its own rules. According to the OECD, Mexico is one of four member countries that still lacks dedicated public funding for the development of AI in government. The tension between the technical capacity being installed and the institutional architecture yet to be built defines Mexico's actual position on the global map of public AI.

Mexico in the Indices: Strength in Adoption, Lag in Governance

In the Latin American Artificial Intelligence Index (ILIA) 2024, produced by the Economic Commission for Latin America and the Caribbean, Mexico ranks sixth regionally with 51.4 points out of 100. Chile leads with 73.07, followed by Brazil (69.30), Uruguay (64.98), Argentina (55.77), and Colombia (52.64). The gap with first place exceeds 21 points.

What the index reveals is not the ranking position itself but the profile that produces it: Mexico shows relative strength in AI research, development, and adoption, but registers marked weakness in the governance dimension. In other words, the country is incorporating the technology faster than it is building the mechanisms to oversee it. That pattern frames the entire reading of what is happening in 2026.

The scale of the challenge is global. According to Gartner projections, by 2026 more than 75% of governments worldwide will use AI to support administrative decisions. The pace of adoption consistently outstrips the pace of regulation in almost every country. What distinguishes one country from another is not whether it is adopting AI, but how far it has progressed in building the rules to govern it.

The Infrastructure Being Installed

The most significant move of 2026 in AI public policy is the Artificial Intelligence Factory, a project of the Agency for Digital Transformation and Telecommunications (ATDT) that contemplates an investment of six billion pesos and is headquartered at the Zacatenco campus of the National Polytechnic Institute. At its core will be the Coatlicue supercomputer, with 15,000 GPUs distributed across 7,500 chassis and the capacity to execute 314 trillion operations per second, which would make it the most powerful in Latin America. Construction began in June 2026; the estimated timeline to full operation is 24 months.

Operational Applications Already in Service

Before the Factory reaches full operation, the ATDT has already deployed several functional systems. Ventanilla 24/7, launched on January 9, 2026, is a virtual assistant that guides citizens in locating services within the gob.mx portal and has served more than 280,000 people. Kuúl is an AI-based multilingual tourism guide. The Semáforo Aduanero prioritizes inspections based on risk analysis. The Graph Model identifies tax evasion networks and irregular invoices in real time.

The Public AI Training Center, operated by INFOTEC and TecNM in partnership with 14 technology companies, aims to graduate 10,000 specialists in 2026 and 25,000 per year in successive cycles. Training areas include cybersecurity, programming, and data analysis applied to the public sector.

What Is Already Happening Inside Government: Adoption Without Protocol

Before these initiatives took institutional form, AI had already penetrated the daily operations of the federal administration through less formal channels. A review of 45 freedom-of-information requests to federal agencies revealed that the SAT uses models to detect shell companies and irregular importers; the SRE operates the PTAT chatbot for consular and passport services; the Ministry of Culture uses image classification and record linking; the SABG normalizes texts with Google Gemini; and ISSSTE and FOVISSSTE manage security events with AI.

In most of these cases, use occurs without mandatory guidelines defining which data is processed, with which models, and under what audit standards. Banco de México is a notable exception: it allocated 8.3 million pesos to AI systems with formal internal guidelines, one of the few cases with documented explicit governance.

Jorge Ordelin, a researcher at the Center for Economic Research and Teaching (CIDE), notes that informal AI use "cannot be systematized or audited," making it impossible to monitor frequency of use and data handling. Juan Manuel Aguilar, of the Center for Research on North America (CISAN-UNAM), warns that uploading official documents to language models entails "indirect data leaks or exposure of sensitive content."

The Regulatory Framework: Between Declarations and Pending Legislation

Between 2022 and 2025, at least six legislative initiatives to regulate AI in Mexico were introduced. None was approved.

On January 29, 2026, SECIHTI and the ATDT presented the Chapultepec Principles, a ten-point framework of ethical guidelines for the use and development of AI. Among its core points: every AI-assisted decision must have an identified human responsible; no decision should be automated if it cannot be explained; and those affected, and to what degree, must be assessed in advance. The principles carry the weight of a declaration, not a binding norm.

On February 11, 2026, Senator Karina Isabel Ruíz Ruíz introduced the National Law to Regulate the Use of Artificial Intelligence. The bill proposes creating a specialized regulatory body with authority to classify AI systems by risk level, issue binding technical guidelines, and impose sanctions. It establishes explicit prohibited uses: fraudulent data collection, ideological profiling, subliminal manipulation, and deepfake creation. It also incorporates the concept of neuro-rights linked to physical and psychological integrity. The proposal is in Senate committees; specialist analysts estimate a low-to-medium probability of passage, given the institutional design challenge and the budgetary impact of creating a new regulatory body.

In parallel, the National Artificial Intelligence Alliance (ANIA) published the National AI Agenda for Mexico 2024-2030 as a coordination reference among public, private, and academic actors, articulating a path of ethical integration without substituting the necessary legal norm.

The Global Map: Where the Leaders Are and What They Built First

Mexico's position becomes clearer in comparative context. The countries topping Oxford Insights' Government AI Readiness Index, including the United States, the United Kingdom, Singapore, and South Korea, combine three elements simultaneously: public computational infrastructure, a national strategy with specific budget allocations, and a regulatory framework with operational supervisory bodies.

In Latin America, the three ILIA leaders illustrate distinct but convergent trajectories. Chile advanced with its National Artificial Intelligence Policy updated in 2024, with explicit funding and clear sectoral mandates. Brazil launched the region's first AI regulatory sandbox in 2025, allowing it to test regulatory frameworks with real cases before legislating at scale. Uruguay combines its digital governance leadership with bilateral agreements to adopt standards derived from the European AI Act.

The shared trait of these three regional leaders is that they built the foundations of governance before, or in parallel with, infrastructure investment. Not after. Armando Zúñiga Salinas, a technology policy analyst, summarizes the structural risk: "Without a national strategy, Mexico risks being merely a user of external technologies," capturing operational efficiency without generating its own strategic value.

Mexico is investing in continental first-tier infrastructure. The window to align the institutional framework with that investment coincides with the Factory's construction timeline: two years.

The Opportunity of the Next 24 Months

The Coatlicue supercomputer will be operational around 2028. That horizon is also a concrete public-policy window: 24 months to build the institutional scaffolding that will allow the infrastructure to operate under accountability standards comparable to those of the regional leaders.

The elements already exist in embryonic form: the Chapultepec Principles as an articulated ethical foundation, the legislative proposal as a technical draft that can be refined with the experience of neighboring countries, the ATDT as an executing agency with operational projects in production, and the AI Training Center as a talent policy already underway. The transition from declaration to binding norm, and from project to institution with mandate and budget, is the central task of that period.

Mexico has assets that few countries in the region can replicate with the same density: a critical mass of technical talent, a diversified industrial base, operational use cases in the federal administration, and a digital agency with demonstrated execution capacity. Integrating those assets under a coherent governance architecture is the step that would convert the infrastructure investment into a durable competitive position on the global map of public AI. That is precisely the type of work where institutional design makes the difference between being a reference point and being a user.

Fuentes:

• https://www.cepal.org/en/pressreleases/latin-american-artificial-intelligence-index-ilia-reconfirms-chile-brazil-and-uruguay
• https://www.infobae.com/mexico/2026/05/05/la-primera-fabrica-de-ia-en-mexico-tendra-la-supercomputadora-mas-poderosa-de-america-latina-y-servira-para-estos-fines/
• https://www.itsitio.com/mx/inteligencia-artificial/gobierno-de-mexico-ya-utiliza-inteligencia-artificial-sin-regulacion-ni-estrategia-clara-advierten-expertos/
• https://basham.com.mx/en/ley-nacional-para-regular-el-uso-de-la-inteligencia-artificial/
• https://www.secihti.mx/ciencia-y-humanidades/principios-de-chapultepec-declaracion-de-etica-y-buenas-practicas-para-el-uso-y-desarrollo-de-la-inteligencia-artificial/
• https://www.excelsior.com.mx/nacional/inteligencia-artificial-eje-estrategico-2026