von JAYANTI DHINGRA
With the prevalence of data migration, there is an increasing threat such data is used against migrants. This means that two types of migration are taking place – migration of people in search for survival and safety (also including the non-refoulement principle), and the other is data migration, which contains personal information of the migrants. This piece argues on the limitations of international law in addressing the issue of relying on machine learning algorithms to control migration.
Increasing Data Migration
The term Migration Governance has been defined by the International Organization for Migration as “frameworks of legal norms, laws and regulations, policies and traditions as well as organizational structures […] and the relevant processes that shape and regulate States’ approaches with regard to migration in all its forms.” This type of governance ultimately imposes authority on migrants through the outputs generated by the algorithmic systems, which in turn becomes a decisive factor on the legal status of a migrant. Some of its examples include iBorderCtrl project or the Eurodac which uses biometric technologies to match the records of migrants.
In 2025, the United Nations High Commissioner for Refugees (UNHCR) estimated that around 120 million people were forcibly displaced from their home territories. However, this raises concerns on the amount of personal data compromised. Such data may include personal identifiers like facial scans, biometrics, financial records, fingerprints/thumb impressions, etc. Further with the advancement of Artificial Intelligence (AI) technologies, the human bias is often assumed to be minimised; however, within such structures, human bias is an inherent part of the system and further amplified by AI. It is used to ensure accurate data interpretation and mapping, detecting errors, etc. It involves predictive systems that analyse the predictability of a person migrating to borders. This raises concerns under international law as the deployment of such AI systems regularly involves non-state actors, which in turn asks questions on the need for state accountability and responsibility while deploying such systems.
Data Migration and the International Law Framework
International law primarily governs the conduct of states. AI deployment is now being increasingly made and sourced by private actors, without directly coming under the purview of international legal obligations. When these systems are created through public-private partnerships, there are no human rights safeguards or accountability mechanisms in place. In fact, this has emerged as a lucrative industry for the private sector.
The International Covenant on Civil and Political Rights (ICCPR), and the International Covenant on Economic, Social and Cultural Rights (ICESCR) impose obligations upon states to protect the rights of non-citizens who are in their jurisdiction. The right to privacy is governed under Article 17 ICCPR and the right to fair trial needs to follow the due process of law (Article 14 ICCPR). The application of these rights to AI driven migration has not been enforced.
The Convention Relating to the Status of Refugees, 1951 was not designed for the digital age. It has no framework on how to deal with algorithmic decision-making. It does not contemplate situations where a person’s refugee status will be decided not by a human, but by a machine learning system trained on historical data fed by humans who also might have racial, nationality or ethnicity biases.
In 2019, the Global Compact for Migration (GCM) was adopted in Marrakesh which specifically addressed international migration. Objective 17 of the GCM requires “eliminating all forms of discrimination” and instead advocates for “evidence based public discourse on migration”. However, due to its non-binding character, it can neither be enforced vis-à-vis other states nor does it hold any legal obligations.
Certain Examples
The below examples illustrate the expanding use of AI at borders for national security and surveillance. For example, iBorderCtrl project, funded by the EU, deploys AI and biometric matching technologies. Similarly, Eurodac, originally designed to record the fingerprints of asylum seekers, has now been expanded to also cover irregular migrants. This depicts a ‘function creep’ where its usage deviates from the original purpose with which it was initially formulated. These systems operate as ‘black boxes’ which makes it difficult to challenge the decision-making process. ‘Project Jetson’ has been introduced by UNHCR to use iris recognition to monitor refugees in countries like Kenya and Jordan. The predictive analytics system is used to assess whether a person is about to attain the migration status. This includes acts like denial of visas, targeted surveillance upon a particular person and based on that person’s behaviour and conduct.
The EU AI Act was adopted in 2024 which will come into force in August 2026. The AI Act classifies migration, asylum and other border control areas to be of ‘high risk’ category mandating more transparency, monitoring, and accuracy testing. However, the AI Act contains a broad national security exemption which implies that its application on border controls gets restricted, allowing for the States to use data driven migration governance without much hesitation by invoking the national security exemption. The United States also have advocated for a more liberal use of AI in migration governance.
Conclusion
While AI is projected to help in achieving meaningful decisions, it comes at the cost of data protection law and international human rights law. In 2023, the United Nations Office of the High Commissioner for Human Rights (OHCHR) issued guidance on how a human rights approach should be considered for digital border governance. However, these documents are non-binding on states, with no legal obligation attached for its non-compliance. AI regulation is fragmented and dispersed, with no data protection law able to account for migrants, who are one of the most vulnerable section of people across countries. There is a need to create a binding international legal instrument governing usages of AI in migration, and it should be able to maintain accuracy over various demographic data in order to avoid racial biasness, while addressing the implementation gaps in the current international framework. Though there are some positive initiatives being taken to include human rights with digital governance, like the EU’s Protect Not Surveil project, it depends on how influential these initiatives and coalitions can create a stronger narrative around the community and the nation at large. It ultimately depends on states whether they use AI for ‘security purposes’ or on advocating human rights of migrants as well.
Zitiervorschlag: Dhingra, Jayanti, Migration Control and its Intersection with International Law, JuWissBlog Nr. 48/2026 v. 27.05.2026, https://www.juwiss.de/48-2026/
Dieses Werk ist unter der Lizenz CC BY-SA 4.0 lizenziert.
