Responsible Use of Artificial Intelligence in Refugee Status Determination
A rights-based policy framework for managing mixed-movement flows in Latin America — Leonardo Delmondes
Rationale
Asylum systems across Latin America are under sustained pressure from large-scale mixed movements, limited institutional capacity, and growing backlogs. AI and digital tools are increasingly used to support information management and triage — but Refugee Status Determination (RSD) is a high-stakes, rights-based process where due process, data protection, and access to effective remedy are essential. Some elements, especially credibility assessment, cannot be automated without undermining fundamental guarantees.
The project starts from three premises: AI should support more efficient claims processing; it should only be used as a bounded decision-support tool; and — based on the author's field experience — a substantial share of existing backlogs could be resolved through administrative proceedings rather than full individual RSD assessment.
Key objectives
Map how asylum systems in selected Latin American contexts currently use AI and manage RSD-relevant information flows.
Identify concrete, rights-compatible AI use cases that support asylum processing without replacing human judgment.
Assess the legal, ethical, and operational risks of AI-supported functions in asylum procedures.
Develop an implementable policy framework defining minimum requirements, risk thresholds, and operational safeguards.
Scope: where AI helps, and where it must not go
Permitted, with safeguards
Country-of-origin information research, retrieval & summarization, with full traceability
Translation assistance, with mandatory human review
Complexity-based prioritization, as non-binding recommendations only
Prohibited
Fully automated final decisions on status or inadmissibility
Automated credibility assessment, emotion analysis, or profiling
Anything that weakens access to explanation, contestation, or appeal
Systems lacking auditability or clear human accountability
Case studies
Peru, Brazil, and Mexico — together accounting for ~17% of the asylum backlog in the Americas in 2025, spanning distinct dynamics from large Venezuelan caseloads to group-based recognition and high procedural pressure from regional mobility.
🇵🇪 Peru🇧🇷 Brazil🇲🇽 Mexico
Main outputs
Policy paper
Analyzing the responsible use of AI in asylum processing, with illustrative case studies from Peru, Brazil, and Mexico.
Policy framework & operational protocol
Minimum requirements, a risk matrix (permitted / conditional / prohibited), and an implementation protocol covering logging, human review, and appeal safeguards.
Where this connects to broader work
Two collective-project contributions extend the same field experience — UNHCR work with asylum authorities in Peru, Mexico, and Ecuador — into wider questions of governance:
Transnational Governance in a Time of Global Disorder
How global disorder and distrust reinforce mixed movements, and how weakened transnational governance limits states' capacity to respond — linking micro-level administrative practice to macro-level governance dynamics.
Modernisation & Mobility Governance
How states use administrative and technological reform to modernize mobility governance, and how "going digital" in a post-AI era blurs the line between efficiency, control, and exclusion.
See a synergy, or want to compare notes?
Always glad to connect with people working on asylum systems, digital governance, or responsible AI in the public sector.