From legal tech to smart governance
Aug 14, 2025
Are we really using tech to fix the government?
Mexico’s “digital government” wave is moving from online forms to AI-driven decision tools. Yet the gap between ambition and impact is widening: connectivity is patchy, algorithms often run as black boxes, and civic participation still lags. This article unpacks three real-world initiatives: Mexico City’s Llave CDMX, Guadalajara’s Visor Urbano, and the SAT’s AI oversight program, and asks whether they advance true smart governance or simply automate old flaws.
Digital governance ≠ Online forms
When policymakers talk about “government tech,” the conversation too often stops at digitising paperwork. True smart governance means opening data, involving citizens in oversight, and making algorithmic rules transparent. Latin America’s largest Spanish-speaking nation offers both promising pilots and sobering gaps: 18 % of Mexican households still lack Internet access, yet AI is already classifying taxpayers by risk
Case Study 1. Llave CDMX & ADIP’s Digital Core
The Agencia Digital de Innovación Pública (ADIP) launched Llave CDMX, a single digital ID that integrates more than 100 public services in Mexico City.
Why it matters
Metric | 2024 Baseline | Goal 2026 |
|---|---|---|
Services available via Llave CDMX | 108 | 200 + |
Avg. time per tramit | 40 min | < 15 min |
Citizen control: residents grant or revoke data-sharing consent inside the portal.
Open source: ADIP publishes code on GitHub, allowing audit and reuse.
Governance test
ADIP built an external “Comité de Transparencia Digital” with NGOs to vet new modules, a rare example of algorithmic accountability in Mexico.
Case Study 2. Visor urbano, Guadalajara
A municipal platform where anyone can check land-use permits, report irregularities, and track zoning changes in real time.
Why it matters
Impact | Data |
|---|---|
Illegal construction reports resolved | 75 % within 10 days (2023) |
Bribes reported by builders | ↓ 45 % vs. 2018 baseline |
Governance test
The code is open-source; neighbouring states (e.g., Baja California) have forked the platform, showing how local pilots can scale nationally when transparency is designed in.
Case Study 3. SAT’s AI for fiscal oversight
Since the 2024 Plan Maestro, the Servicio de Administración Tributaria has used machine-learning graph analytics to flag high-risk taxpayers and detect synthetic invoices.
Why it matters
Impact | Data |
|---|---|
Additional revenue attributed to AI flags (2024 Q1) | MXN 18 bn |
Time to identify fraudulent CFDIs | ↓ 70 % |
Governance test
The SAT publishes zero technical documentation on model bias or error rates. Without public audit, the tool risks automating false positives that disproportionately hit SMEs.
The persistent digital divide
Despite flagship projects, basic access remains uneven:
Connectivity: 18 % of households lack Internet; rural broadband coverage below 60%.
Skills: Only 30% of public servants report “intermediate” digital capability in an OECD assessment (2023).
When AI systems score credit worthiness or tax risk in such a context, they can replicate offline inequality at algorithmic speed.
“If algorithmic tools are built without meaningful oversight, we swap human discretion for machine opacity—and citizens still can’t appeal decisions they don’t understand.”
— Mariana Salazar Albornoz, Chair, Ibero-American Data Protection Network (interview in El Economista, 14 May 2025)
Smart governance is a process, not a portal. Without hard guarantees; open algorithms, civic oversight, universal connectivity, “digital government” risks automating yesterday’s inefficiencies.
Some questions to reflect on.
Will the upcoming federal Ley General de Ciberseguridad mandate algorithmic audits?Can local successes like Visor Urbano survive political turnover?How will Mexico bridge its rural digital divide before scaling AI-driven public services?