Peru - National AI Strategy

Overview

The "Estrategia Nacional de Inteligencia Artificial (ENIA) — Documento de Trabajo para la Participación de la Ciudadanía 2021–2026" is a draft strategy led by the Presidency of the Council of Ministers, Secretaría de Gobierno y Transformación Digital (PCM-SGTD). It was issued as a public working paper to gather inputs from citizens, academia, industry and civil society and to align AI development with the national Digital Transformation agenda established by Decreto Legislativo N° 1412 and its Reglamento (DS N° 029-2021-PCM). The draft sets out a five-year horizon (2021–2026) of objectives, axes of action and proposed measures intended to promote inclusive, ethical and transparent AI in Peru, while strengthening public sector use and national capacities. The ENIA draft is available for download and consultation from the Peruvian government portals (see the official publication page and the working PDF published by PCM: ENIA — PCM page and ENIA — working document (PDF)).

Definitions

The ENIA draft establishes working definitions to ensure consistency across policy actions. It defines "artificial intelligence" broadly as a set of techniques and systems that enable machines to perform tasks that would normally require human intelligence, including machine learning, statistical inference, knowledge-based systems, and related data-processing algorithms. The document distinguishes between "AI systems" (specific deployed systems and models), "models" (trained algorithms), "datasets" (data collections used for training and inference), "AI governance" (institutional, normative and operational frameworks) and "human-centered AI" (design and deployment practices that prioritize human rights, non-discrimination and explainability). It also articulates definitions for "impact assessment" (evaluations of potential harms/benefits), "transparency measures" (documentation and registries), and "high-risk" uses (applications with significant potential to affect rights or fundamental services).

Governance and Institutional Framework

ENIA proposes a multi-level governance architecture anchored at the PCM through the Secretaría de Gobierno y Transformación Digital as the coordinating body for AI policy. The draft recommends creating an inter-institutional advisory committee composed of public sector agencies, the national data protection authority, sectoral regulators (health, education, justice, finance), academia, civil society and private sector representatives. This structure is intended to provide policy direction, prioritize use-cases, and oversee standards development. ENIA also suggests operational subgroups for data governance, ethics and impact assessment, talent and R&D coordination, and public procurement of AI solutions. The governance design intends to leverage existing government platforms (e.g., Datos Perú / Plataforma Nacional de Datos Abiertos) and legal instruments such as the Government Digital Law (Decreto Legislativo N° 1412). For coordination and capacity building the ENIA draft suggests links and collaboration mechanisms with the national data protection authority (Autoridad Nacional de Protección de Datos Personales) and international organizations; for reference see the PCM transformation digital portal: PCM — Transformación Digital and the ANPD portal: ANPD — Data Protection Authority.

Key Focus Areas

ENIA structures its actions through several focus areas and proposed measures. First, data governance: promote open, documented and interoperable public datasets, metadata standards, and data quality controls to feed reliable AI models while respecting privacy and data-protection law. Second, capacity and talent: invest in training, scholarships, and R&D infrastructure (including cloud and high-performance computing) to build national capabilities. Third, public sector adoption: identify priority use-cases (e.g., health triage, tax analytics, service delivery), run pilots in controlled environments, and develop procurement guidelines and pre-contractual evaluative tools. Fourth, ethics and rights: adopt human-centered principles (fairness, non-discrimination, inclusion, explainability) and require ex-ante impact assessments for higher-risk public deployments. Fifth, risk management and cybersecurity: define minimum-security standards for models, incident reporting procedures and contingency mechanisms. Sixth, standards and certification: promote technical standards, model documentation (model cards/datasheets), and voluntary certification schemes that can later be formalized into regulatory requirements. Finally, international cooperation: align with OECD, UNESCO and regional AI initiatives for interoperability and good practice adoption.

Implementation Framework

The draft recommends a phased, iterative implementation approach combining policy development, pilots and capacity building. Phase 1 (initial; 2021–2022) focuses on institution building, stakeholder engagement, mapping of public use-cases and development of guidelines for procurement and impact assessment. Phase 2 (scale-up; 2022–2024) expands public pilots, supports R&D investments and establishes registries/documentation standards. Phase 3 (consolidation; 2024–2026) aims to mainstream practices, evaluate outcomes and propose regulatory instruments as needed. The ENIA envisions coordination instruments such as an AI policy roadmap, annual monitoring reports to Congress and public dashboards summarizing AI deployments. The draft proposes funding lines via public budgets and partnerships with development banks and international cooperation agencies to accelerate deployment and training initiatives.

Monitoring and Evaluation

Monitoring is framed as a continuous process with periodic public reporting. ENIA proposes quantitative and qualitative indicators: number of trained professionals, AI research outputs, number of vetted public pilots, percentage of public services improved by AI, data quality metrics and compliance with documentation standards. The draft suggests an annual public report on progress and an open dashboard with metrics driven by the PCM secretariat. Additionally, the ENIA draft recommends independent evaluations of major deployments and creation of a feedback channel for citizens to report harms or poor outcomes. These mechanisms are intended to inform adjustments, re-prioritization and to foster transparency.

Penalties, Liability, and Appeals

As a working document, ENIA does not itself establish civil or administrative penalties; instead it highlights the need to harmonize future liability and redress frameworks with existing legal regimes (e.g., data protection law, consumer protection, sectoral regulation). The draft advises developing guidelines for accountability and incident management, including complaint mechanisms and administrative redress routes for harms caused by public-sector AI. It also recommends mapping existing liability gaps and proposing legislative or regulatory solutions — for example mandatory impact assessments for high-risk systems and potential certification requirements — to be refined through follow-up regulation. For data-protection-related harms, the draft points to the remit of the ANPD and sectoral supervisors.

Relationship to Other Instruments

ENIA is designed to sit within the broader national digital policy architecture: it complements the Estrategia Nacional de Gobierno de Datos, the Estrategia Nacional de Seguridad y Confianza Digital, the Ley de Gobierno Digital (Decreto Legislativo N° 1412) and its Reglamento (DS N° 029-2021-PCM). It also anticipates alignment with sectoral laws (health, education, public procurement, judicial administration) and with data protection obligations under Peru's data protection regime (Law N° 29733 and its implementing rules) and the ANPD's oversight activities. The draft explicitly recommends coherence with international instruments and technical standards to facilitate interoperability and cross-border data flows under safe conditions.

International Alignment

ENIA emphasizes alignment with international principles and standards, citing recommendations and guidance from the OECD (AI Principles and Recommendation), UNESCO (Recommendation on the Ethics of AI), and regional initiatives in Latin America. The document advocates active participation in multilateral dialogues, technical standardization bodies (ISO/IEC), and regional public-sector cooperation to exchange good practices for public-service AI. It also proposes partnerships with multilateral banks and the OECD/CAF network for technical assistance and funding to implement priority actions.

Implementation Timeline

Sources and References