Entering the Algorithmic Age of EU Migration Governance

At a glance

The European Union’s long-negotiated Pact on Migration and Asylum is set to be implemented from June 2026.

AI-powered technologies have become central to the bloc’s asylum and migration governance, promising greater efficiency, predictive capabilities, and improved self-reliance among asylum seekers and migrants.

Yet, in the context of slow-moving legislation and the rising influence of public-private partnerships, questions remain regarding the legal, social and ethical risks of such technologies.

Ridgeway analysis

In Brussels, all eyes are on the EU’s Pact on Migration and Asylum. The Pact, set to be implemented from 12 June, represents one of the most substantial migration reform projectsin EU history. With Member States scrambling to prepare for the Pact’s implementation deadline, many operational and political challenges remain. The Pact itself has ten ‘building blocks’, outlined in the Common Implementation Plan. Against each building block, there are a number of required measuresfor Member States to ensure consistent and adequate implementation procedures, legislation, capacity and resource.

The Pact’s final push comes as AI technologies becomes more deeply embedded in migration governance. Spurred on by the Covid-19 pandemic, geopolitical instability and ‘migration crises’, governments are welcoming these new technologies with open armsin order to manage refugee and migration influxes.

At the EU-level, the Pact presents an opportunity to leverage digital tools to relieve pressure on overstretched national asylum and migration systems. New technologies are set to play a key role in various areas, including decision-making at the border, relocation of asylum seekers and refugees, and return and reintegration procedures. For example, in March the European Parliament voted in favour of adopting the new EU Return Regulation, which significantly expands digital surveillance infrastructure – including through mass data collection, data transfers to third countries, and technology-facilitated risk assessments and profiling.

Developments in digitised border control, surveillance and management bring existential challengesfor the EU – many of which are invisible by design. Meanwhile, public-private partnerships are playing a starring role, with private companiesincreasingly influencing policy on migration governance and often offering militarygrade resources. As legislation struggles to keep up, asylum and migration governance is afflicted with more and more dividing lines – between countries, and among migrants themselves.

Lagging legislation

Understanding of the use of AI-powered technologies in asylum and migration management remains highly limited, while the development of relevant legislation is slow-paced. To help fill these gaps, the EU Agency for Fundamental Rights (FRA) has embarked on a research project to examine how current AI technologies comply with fundamental rights, with the aim of providing practical guidance to the bloc’s national authorities. The legislation informing this research will largely be the EU AI Act (a key reference point for the Pact), EU Charter of Fundamental Rights, and EU data protection legislation. Alongside this existing legislation, in October 2025 the European Commission also launched its Apply AI Strategy to complement the AI Act and proposed Cloud and AI development act. Although a turning point in EU technology policy, questions remain around the strategy’s undetermined budget and whether it can meet sector-specific needs.

Beyond EU legislation, in September 2024 the world’s first international (and critically, legally binding) treaty on AI opened for signature. The Framework Convention on AI and Human Rights, Democracy, and the Rule of Law aims to ensure AI systems are governed in a rights-based, inclusive and globally connected way. While remaining ‘technology-neutral’ to avoid only targeting specific, current technologies that may fade overtime, the framework is attempting to plug legal gaps as technology advances at pace.

Despite these efforts, legislation is racing to keep pace with rights’ concerns. The #ProtectNotSurveil coalition is one group concerned about the regulation of digital surveillance in migration management. In particular, the coalition has criticised the AI Act for not going far enough, calling for bans on “harmful AI practices” – such as automated profiling systems and remote biometric identification. Meanwhile, a joint statement by 13 rights-based groups recently called for the AI Act to expand the safeguarding applicability in Article 83 to include all major EU migration databases – such as Eurodac, the Schengen Information System (SIS) and European Travel Information and Authorisation System (ETIAS).

Amid these calls, in March the Council of Europe Division on Migration and Refugees (DMR) made several updates to its reference document on non-binding legal instruments regarding migration and refugees. The document notesthe “transformative role” of AI in migration management, both for operational efficiency and service accessibility among migrants, refugees and asylum seekers. New recommendations have been noted, however, on risks for fundamental rights protection, with Member States urged to ensure AI developments adhered to ethical standards, human rights safeguards, and non-discriminatory practices. As the new Pact comes into force, Member States will be expected to establish impact assessments and independent oversight bodies to manage AI technologies.

Private sector influence

Public-private partnerships are becoming a defining feature of recent AI transformations in asylum and migration governance. More and more, private technology companies play the role of designer, contractor and operator for border infrastructure, and so often influence how migration is managed, monitored and perceived. The Mixed Migration Centre recently highlighted a “perverse circular economy” where private companies can benefit twice – firstly, when providing technologies to the defence industry (enabling activities which may lead to mass displacement) and, secondly, when providing technologies that help manage the consequent population movements. And while many governments around the world are working to contain migration, many private actors are instead seeing opportunity. Indeed, EY recently described migration infrastructure as the “next competitive edge” in the context of widescale labour shortages, noting opportunity in the convergence of demographic, geopolitical, climate and technological changes.

In Europe, the EU’s border agency Frontex is increasingly outsourcing support on AI development and biometric surveillance infrastructure to IT and defence companies. Europol is also deepening – reportedly unchecked – ties with the private sector, including work to tackle migrant smuggling and human trafficking.

Across the Atlantic, companies such as Palantir Technologies, Microsoft, Clearview AI, Anduril and Amazon are central to US immigration enforcement systems – as documented in Petra Molnar’s book that explores experimental border surveillance technologies. Even humanitarian actors are now tapping into private sector technologies. For example, the UN’s World Food Programme and Palantir launched a five-year partnership in 2019 to help streamline relief operationsthrough an AI and machine learning software platform, while the International Organization for Migration (IOM) has contracted private companiesto expedite migrant removal processes.

Partnerships with private companies come with dependencies and risks. Companies are often not held to the same legal or accountability standards as multilateral institutions or governments. For example, as the new Pact comes into play, private companies deploying AI systems are not bound by the same legislation on technology use, with the EU AI Act currently only encouraging companiesto make “voluntary commitments”.

Questions around responsibility, data ownership and security have also become murkier. Private companies are responsible for greater amounts of sensitive data (often concerning individuals with protection needs), while data incidents can create confusion around liability between the state and contractor. Partnerships between humanitarian and private actors have further been subject to questions around data protection and human rights considerations.

Deepening divides

The testing, piloting and deployment of AI-powered technologiesin asylum and migration management varies across Member States. Such technologies require significant state capacity, including in infrastructure, energy, institutional expertise and trained staff. Countriesfurther ahead include Austria, Belgium, Bulgaria, Denmark, Estonia, Germany, Greece, Hungary, Ireland, Latvia, the Netherlands, and Sweden. Others are at earlier stages or rely on fragmented or outdated systems. With the pact promising to standardise procedures across the bloc, this variation has become challenging – for states and migrants alike.

For example, since 2017 Germany has used an AI language detection tool to support asylum procedures. Inspired, the EU Agency for Asylum recently launched a 10-year project to create an EU-wide platform that would identify asylum applicants’ country of origin (COI) through AI language detection tools and language analysts. Under the new pact, the determination of COI will likely place during the pre-entry screening. If language detection tools are used at this stage under the new rules, applicants from certain nationalities may face heightened risks of unfair decisions or significant barriers when lodging and challenging asylum claims.

At the same time, uneven technological progress across Member States meansinteroperability and governance challenges. Particularly under the new pact, EU asylum and migration governance increasingly relies on interconnected databases and digitalised systems. Disparities in institutional readiness and technical capacity may create implementation failures, delays in border and asylum procedures, and increased political pressure from EU institutions. Counter to the Pact’s principle of “burden-sharing”, more technologically advanced countries – such as Germany – may be disproportionately relied upon.

Asylum seekers, refugees and migrants are navigating these disparate and complex digitalised systems. According to interview insightsshared by the Migrants’ Rights Network, asylum seekers generally support the use of AI to standardise or speed up decision-making, but believe final decisionsshould be made by a human.

Despite limited publicly available data on the real-life effects of AI, there are concerns that existing inequalities may be reinforced through digital exclusion, ethnic and racial bias, and accessibility barriersfor people with disabilities. In asylum decision procedures, key risks already identified include:

• discrimination and biased algorithmic profiling;

• inaccurate or unfair risk assessments;

• opaque surveillance systems;

• and limitations on AI’s ability to interpret human behaviours and cultural complexities;

AI-centred approaches to relocation efforts offer some insight into these complexities. Relocation is a key part of the Pact’s solidarity mechanism, which essentially encourages greater responsibility sharing between Member Statesto support countries under “migratory pressure”. An emerging trend in relocation is algorithmic matching processes between asylum seekers or refugees and host communities. The US has already successfully used a machine-learning software called Annie™ MOORE to resettle at least 1,000 refugees, informed by historical relocation data on employment outcomes. The US and Switzerland have further used a similar tool called Rematch (expected to improve refugee employment outcomes by up to 70%), while Ukrainian refugees are being relocated across Germany through the Berlin Governance Platform’s Re:Match programme that usesthe Pairity matching algorithm which is based on preference-ranking and biographical data.

Download

A downloadable PDF of this Briefing can be accessed here.

For more information, questions, or to find out how Ridgeway can support your organisation's research requirements, contact us at info@ridgeway-information.com.