As reflected in the New Pact on Migration and Asylum, which is set to enter into force this year, the European Union increasingly frames human mobility through the lens of crisis management and the protection of its citizens against an external threat. This threat is often constructed as a racialized “Other”, embodied by the figure of the migrant (https://picum.org/blog/open-letter-eu-human-rights-risks-migration-pact/). This framing legitimizes the expansion of surveillance infrastructures and exclusionary policies that risk undermining the fundamental rights of people on the move, reinforcing the expansion of “Fortress Europe”, where mobility is restricted based on class, race, gender, origin, and especially nationality (Engelbert et al., 2019).

This logic is operationalized through multiple, interrelated forms of externalization. First, responsibility for asylum is increasingly shifting towards peripheral Member States under the Dublin III Regulation (2013) or even extended beyond EU borders through cooperation with third countries via frameworks such as the Asylum and Migration Management Regulation (EU) 2024/1351 (AMMR), which legally formalizes the importance of cooperation with third countries as a fundamental element of EU migration and asylum policies (https://eumigrationlawblog.eu/cooperation-with-third-countries-within-the-eu-legislative-reform-on-migration-and-asylum/). Second, operational responsibilities are increasingly delegated to EU agencies such as Frontex, as well as to private actors, raising concerns regarding their limited accountability and profit-driven incentives. As noted by Guiraudon and Lahav (2000), states have long sought to evade legal and human rights constraints by displacing migration control “upwards, […] downward […] and outward”.

This outsourcing of border management responsibilities has been accompanied by a growing reliance on algorithmic systems designed to categorize and monitor people on the move. Increasingly deployed within migration management procedures financed by the European Union, these technologies exemplify what Madianou (2025) conceptualizes as “technocolonialism”. Using a colonial lens helps to illuminate the power asymmetries embedded in such practices. For instance, meaningful consent is effectively absent in contexts such as refugee camps, where refusal to provide biometric data may result in denial of access to essential aid and shelter. Similarly, digital tools such as chatbots, which are often trained primarily in English, can impose Eurocentric assumptions about decision-making on populations whose lived realities differ significantly from the context where these tools were trained (Madianou, 2025). When looking at the increasing emphasis of EU border policies on the use of AI with “technocolonialism” lenses, it is clear how biometric technologies that are presented as efficiency tools are actually further reproducing structural discrimination. By embedding normative assumptions of “whiteness” into their design, they are transforming existing forms of inequality in objective facts (Browne, 2015).

Against this sociological backdrop, this Special Issue examines the growing deployment of algorithmic tools across the external and “pre-frontier” spaces of the European Union, and throughout different stages of migration and asylum procedures. As documented by a report of the Algorithmic Fairness for Asylum Seekers project (2023), artificial intelligence is already widely used in this field. Applications include forecasting migration flows; conducting risk assessments and profiling; processing visa, travel authorization, and citizenship applications; verifying identity and detecting fraud through tools such as emotion recognition and mobile phone data extraction; categorizing applicants based on perceived risk; enabling electronic monitoring; allocating welfare benefits; and matching individuals to their places of residence. These technologies are embedded within large-scale EU data infrastructures, such as the European Asylum Dactyloscopy Database (Eurodac), which collects and compares biometric data and can influence the assessment of protection claims. This is raising concerns for the compliance with fairness principles in asylum decision-making.

Under EU law, asylum decisions must be taken on an individual basis, a requirement that sits uneasily with AI systems that rely on patterns derived from historical data. Moreover, one central criterion of fair asylum procedures according to EU law is the assessment of a “well-founded fear” of persecution by the asylum seeker, which requires a type of reasoning that is forward-looking and goes beyond the inductive reasoning of AI technologies. The use of AI may also undermine procedural safeguards, including the principle of shared burden of proof between applicants and authorities (Article 4(1) of Directive 2011/95/EU) and the principle of the benefit of the doubt (Article 4(5)), both of which are designed to ensure that claims can be fairly assessed even in the absence of extensive documentary evidence (Dumbrava, 2025).

The use of these technologies is not confined to EU Member States, but they are increasingly deployed in non-EU countries. A report by the Translational Institute (2025) highlights the deployment of these technologies in Africa within the Africa-Frontex Intelligence Community (AFIC), a network of 31 African states with the aim to exchange biometric data, operational practices, and risk analysis related to border management (Gkliati & Kilpatrick, 2025). Many projects associated with the AFIC involve the provision of surveillance equipment or the training of local police. As noted by the report, this project-based collaboration often falls outside formal Frontex working arrangements, thereby circumventing oversight by the European Parliament and existing human rights safeguards. Within these projects, the EU funds and exports intrusive technologies, including surveillance drones and biometric identification systems, enabling third countries to monitor migration routes and populations. The expansion of technological surveillance beyond EU borders feeds into the idea of “technocolonialism” and the risks resulting from such policies. In 2022, for example, a former Spanish intelligence agent revealed that technologies supplied by Spain had been used by authorities in several African countries to target opposition groups and activists (Popoviciu, 2023).

Within the EU, the use of AI has not been promoted without thinking about the consequences and designing frameworks to safeguard people’s privacy. For this purpose, the EU AI Act has been developed as an innovative instrument. However, the Act establishes a distinct regime for AI systems used in law enforcement, migration control, and national security, which introduces broad exemptions for the use of these technologies on “othered” bodies. Between 2014 and 2022, the EU allocated over €250 million to 49 projects focused on the development of border technologies (PICUM, 2024). Ongoing initiatives illustrate the EU’s efforts to expand the technocolonial “Fortress”, such as the ROBORDER project, which aims to develop fully autonomous border surveillance systems using robotic platforms across air, land, and sea domains (https://roborder.eu/the-project/aims-objectives/); ODYSSEUS, which seeks to enhance border control processes through biometric technologies (https://odysseusproject.eu/); and Promenade, which focuses on advanced data analytics, including pattern detection, risk assessment, and predictive modeling in maritime contexts (https://www.promenade-project.eu/about/).

Given the increasing enthusiasm for testing these systems within and beyond the EU territory, this Special Issue explores the role, impact, and implications of AI and automated decision-making in the migration and asylum domains, with particular attention to their practical implementation in different national contexts. Rather than treating AI as a neutral technical innovation, this Special Issue approaches these systems as socio-legal constructs that reshape decision-making processes, redistribute discretion, and potentially recalibrate the balance between administrative efficiency and the protection of fundamental rights.

To create an overview of the current state of the application of digital tools and AI within national migration governance, several authors across the European Student Think Tank have written the following articles. They include the presentation of the current national context on migration policy and governance, highlighting challenges and recent developments regarding the use of AI. Further, the authors outline which AI tools and actors are involved, the outcomes of the policy implementation, and how the application of AI in different areas of policy development and implementation is framed. Based on this analysis, the implications on human rights are being explored, and the authors develop policy recommendations at the national and/or EU level.

Ilia Torkanievskyi focuses his analysis on Lithuania, which, after a slow decline in population after 1991, is experiencing an increase in immigration since 2019, mostly due to the Russian war against Ukraine and Belarus’ push of asylum seekers towards EU countries. The recent developments were met with a framing of migration as a security issue, leading to the expansion of surveillance and border control. Within this context, Lithuania adopted automated decision-making systems to centralize procedures and created a single platform to digitally regulate all migration-related administration. Torkanievskyi evaluates how AI tools impact the administrative procedures, with a focus on the Migration Information System (MIGRIS) platform and the use of AI-assisted surveillance technology at the Lithuania/Belarus border.

Due to the Syrian Civil War, other violent regional conflicts, and unique treaties with European countries, Türkiye is the host of one of the largest numbers of refugees worldwide. Hayrünnisa Çelik analyses Türkiye’s migration governance, its shift from emergency reception to long-term arrangements and the role of digital tools and artificial intelligence. Despite the lack of publications regarding evaluations of the implementation of AI within migration governance, Çelik examines how those tools are applied by governmental authorities and the consequences for refugees and irregular migrants. She particularly highlights the use of electronic databases for internal migration governance and the use of surveillance tools as part of border management.

Alexandra Sima presents the application of digital tools and AI within the French migration governance, with a particular focus on asylum seekers. With ongoing violent conflicts in former colonial occupied territories such as Guinea, the Democratic Republic of Congo, and Côte d’Ivoire, as well as Ukraine, France has experienced an increase in asylum applications. Given the increasing pressure on its bureaucratic system, France is increasingly relying on EU-wide databases and cooperation, and aims to adopt more digital tools and AI-supported systems into its migration governance and administrative processes. She highlights the use of Eurodac, the biometric database for fingerprints of asylum seekers and irregular migrants, and evaluates the intended expansion of Eurodac as well as the cooperation with Frontex. Sima’s article includes concerns regarding the transparency of automated processes and the use of AI, as well as an in-depth analysis of their impact on human rights and their potential misalignment with EU law and values.

Considering Cyprus within the context of its relatively high rate of asylum applications and other migrant communities, Aagya Tripathee provides an overview of the role of digital tools in Cyprus’ migration governance. Tripathee analyzes Cyprus’ cooperation with Frontex and Greece on surveillance projects in the Mediterranean and the usage of European databases, while also evaluating the current process of implementing more digital and artificial tools, both in administration and physical border control. She further incorporates the human rights violations and legal rulings that Cyprus faced in the context of its migration governance into her analysis. Highlighting both the pressure and limited resources of Cyprus, as well as its human rights violations and breaches of EU law, Tripathee outlines recommendations on the further implementation of digital tools.

Located on the Mediterranean, Italy is one of the EU countries that refugees and asylum seekers access the most. Flavia Onwuelo analyzes the increased use of digital tools and AI in Italy’s migration governance, considering the context of the large number of asylum seekers and the shift in migration policy under the Meloni-led government, including the cooperation with Albania. She includes both the national deployment of AI tools and Italy’s cooperation with EU-wide databases in her analysis, before analysing the impact on human rights and compliance with international agreements. Based on her analysis, Onwuelo presents policy recommendations for both the national and EU level.

In 2015, Merkel famously stated “Wir schaffen das” (“We can do this”), introducing the framework to welcome and integrate refugees and asylum seekers into Germany. The framework emphasized Germany’s historic responsibility to protect human rights and aligned with the Christian values of altruism. Tamaya Sauerwein explores how this approach to migration governance has shifted in the past years, as the formerly Merkel-led Christian Democratic Party (CDU) is now aligning its language, narratives, and policies with those of far-right movements. Sauerwein explores how the modernization and digitalization of migration governance is applying AI, and how Automated Decision-Making Systems are employed at different levels of German migration governance, including speech analysis softwares, “phone scrapings”, and federal databanks. Sauerwein further explores the consequences on human rights, as human bias is inbuilt in software, which in turn cannot be held accountable for discrimination or the damages caused. 

As the outline of the chapters previews, there is an ongoing effort by authorities to integrate technology and AI into national and supranational migration governance. This tendency is accompanied by growing concerns for human rights, transparency, and accountability. Justified with arguments of “efficiency”, the integration of technological tools seems to align with anti-immigration policies of conservative and far-right parties/groups, aiming to reduce a certain form of immigration to European countries. As reflected in the chapters, the application of AI and technology focuses on refugees, asylum seekers, and other migrants who lack financial and legal resources, hold certain citizenships, or belong to certain religious or ethnic groups. The authors recognize that European countries are experiencing a strong shift to the far-right, which aims to reduce immigration of people of color, Muslims, and people with scarce financial resources. At the same time, many governments are determined to attract people with certain educational backgrounds or investment perspectives to strengthen their workforce. Focusing on their own economic interests, European countries enable controlled migration flows, draining talent and labor along global supply chains of human capital.

This phenomenon highlights that immigration policies focus increasingly on profit rather than human rights considerations and that educational and financial capital enable cross-border mobility, leaving those who have those opportunities denied excluded from legal migration routes.  Therefore, it should be noted how research and public discourse focus nearly exclusively on the application of AI and digital technologies to control and restrict the mobility of people who seek asylum, safety, or opportunities of economic survival. 

The chapters of this Special Issue reflect these rapid developments in migration governance and provide an excellent overview of national and European interaction, policies, and the application of digital tools for migration governance.

In the name of the EST Working Group on Migration,

Chiara Todesco & Helena Bahr

References

Algorithmic Fairness for Asylum Seekers and Refugees (AFAR) Project. (2023). Interim report: October 2022 – November 2023. Hertie School, Centre for Fundamental Rights. https://www.hertie-school.org/fileadmin/2_Research/1_About_our_research/2_Research_centres/Centre_for_Fundamental_Rights/AFAR/Reports/AFAR_-_Interim_Report__2023_-_Website.pdf. 

Browne, S. (2015). Dark matters: On the surveillance of Blackness. Duke University Press. 

Dumbrava, C. (2025, July). Artificial intelligence in asylum procedures in the EU (PE 775.861). European Parliamentary Research Service. 

Engelbert, J., Awad Cherit, I., & van Sterkenburg, J. (2019). Everyday practices and the (un)making of ‘Fortress Europe’: Introduction to the special issue. European Journal of Cultural Studies, 22(2), 133-143. 

Gkliati, M., & Kilpatrick, J. (2025, July). Exporting borders: Frontex and the expansion of Fortress Europe in West Africa. Transnational Institute. https://www.tni.org/files/2025-07/Exporting-Borders-Report-web.pdf. 

Guiraudon, V., & Lahav, G. (2000). A reappraisal of the state sovereignty debate: The case of migration control. Comparative Political Studies, 33(2), 163–195. 

Madianou, M. (2025). Technocolonialism: When technology for good is harmful. Polity Press, 256.

Platform for International Cooperation on Undocumented Migrants (PICUM). (2024, April 4). A dangerous precedent: How the EU AI Act fails migrants and people on the move. https://picum.org/blog/a-dangerous-precedent-how-the-eu-ai-act-fails-migrants-and-people-on-the-move/. 

Popoviciu, A. (2023, July 26). How Europe outsourced border enforcement to Africa. In These Times. https://inthesetimes.com/article/europe-militarize-africa-senegal-borders-anti-migration-surveillance.

Lithuania

Written by Ilia Torkanievskyi

Edited by Kristina Welsch

National Migration Context

Lithuania provides an example of integrating AI technology into asylum and migration management: The rapid development of artificial intelligence in 2020, along with significant migration and demographic pressures, forced the government to expand its digital systems and automated processes to improve migration management. 

After Lithuania regained its independence in 1990, its migration landscape has changed dramatically. Since then, emigration and low birth rates have caused a slow decline in population growth, leading to the country’s population decrease of 20% in its first two decades of independence. Finally, 2019 marked the first year  that Lithuania experienced a positive net migration (+10,794) after decades of demographic decline (European Migration Network, 2025).

The most recent migration environment in Lithuania has been strongly influenced by two factors: the Russian full-scale invasion in Ukraine and the weaponisation of migrants by Belarus. First, the mass arrival of refugees from Ukraine in 2022 created huge administrative challenges for the state. Ever since the Russian full-scale invasion in Ukraine, Lithuania has received more than 100,000 Ukrainian refugees, putting a constraint on the migration system (Official Statistics Portal, 2026). The second important development was the migration crisis presumably orchestrated by Belarus in 2021. Lithuanian officials stated that the Belarusian government had deliberately organised and pushed irregular flows of migrants toward the European Union, using them as a tool of political leverage (Radio Liberty, 2022). 

Both these events changed the way Lithuania’s government approaches migration management. While migration was considered  a mainly administrative issue in the past, it is now framed as a matter of national security (Permanent Mission of Lithuania to the United Nations, 2025). Therefore, the government is intensifying border controls, building barriers along the border with Belarus and expanding the use of surveillance technology. Simultaneously, the Migration Department is experiencing growing pressure because of the drastic increase in applications for residence permits and migration services. The number of applications has increased several times, from 5,178 in 2015 to 54,000 in 2024 (Prague Process, 2026). This has created delays and increased the workload for migration service employees even more.

In an attempt to tackle these issues, Lithuania has chosen a path of digitalisation and the adoption of automated decision-making systems. Thus, in 2019, the Lithuanian Migration Information System (MIGRIS) was introduced as the main platform for all migration-related work (MIGRIS, 2020). This platform centralised migration-related administrative procedures and has been established as the only system through which applicants can book appointments for residence permits and visas (VFS Global, 2026). These factors have contributed to the need for and wider usage of automated tools and AI decision-making systems within the migration administration. 

The Use of AI in Asylum and Migration

As mentioned above, MIGRIS, being the main platform for Lithuania’s migration management, has become a very advanced system that includes automated decision-making. The system is also enabled to review submitted documents automatically. Such revision also includes pulling information from both national and international databases, including the Visa Information System (VIS) and the Schengen Information System (SIS) (MIGRIS, 2020). Consequently, any application for migration became more organised, because government officials do not have to manually verify identity, check a person’s background or look for potential security concerns (Fathallah, 2026). On top of that, MIGRIS has automated other processes. For instance, the platform can verify whether the applicant meets financial requirements, i.e. providing proof of being able to sustain themselves, by analysing bank statements or other related documents (MIGRIS, 2020). Another example of an automated decision-making process is the system’s management of labour migration quotas. These quotas are measured based on the country’s deficit of workers and the staff shortages across different sectors. Thus, once the algorithm decides that quotas are fulfilled, the procedure gets more “strict” and sends applicants for additional review. The usage of automated decision-making has become even more prominent in 2025, when Lithuania reduced its labour migration quotas from 40,250 in 2024 to 24,830 in 2025 (Langeveld, 2025). This, in turn, has created a demand for more efficient systems for processing and filtering applications, because the system is now prioritising high-skilled workers. 

Even though MIGRIS is positioned as a decision-support tool, it is actually one of the most significant devices impacting migration decisions. If applications fail to meet predetermined criteria, such as past legal checks or financial statements, they are filtered before they can reach a human member of staff. Thus, automated screening also impacts which cases are examined by migration officials and which are flagged as rejected (European Commission, 2022). 

Automated systems are also widely used by the State Guard Service (SGS), following the 2021 migration crisis on the border with Belarus. Since then, the SGS have implemented AI-assisted surveillance technologies, including drones with thermal cameras and optical sensors, to monitor the 680-kilometre border with Belarus (Rees, 2022). Artificial intelligence plays an important role in border monitoring, as it recognises a person’s movement and is able to  distinguish it from an animal’s, sending a signal to border control in case illegal border crossing is detected. This technology is the earliest stage of automated decision-making, or rather, enforcement, in the migration process. 

Human Rights Implications

While automated systems and AI simplify the process and the staff’s lives, they also lead to several human rights concerns. One of the most important issues that automated technologies might overlook relates to the principle of non-refoulement, which is a part of international human rights and refugee law prohibiting a state from returning people to the countries where they can face persecution, serious harm or danger (Protecting Rights at Borders, 2023).

The AI-assisted surveillance systems on the borders, for example, cannot assess the person’s personal circumstances (Vavoula, 2021). Once the border guards’ reaction is triggered, their responses usually result in pushbacks, which do not give migrants a chance to gain access to asylum procedures. Therefore, if asylum-seekers are turned away before their claims can even be assessed, their right to seek international protection is effectively denied (The Seimas Ombudsmen’s Office, 2021).

Another important issue concerns data protection and privacy. Personal data, including biometric information like fingerprints and facial recognition, is analysed and collected by the same platform – MIGRIS (State Data Protection Inspectorate, 2023). Then, this collected personal data is usually stored in large European databases like Eurodac or the Visa Information System. These databases were initially intended to foster cooperation between the European Union member states, but they also raise questions about the storage, sharing and protection of personal data (European Commission, 2022). Asylum seekers are especially vulnerable in this context because if their data is shared with other authorities or countries, they risk being exposed to multiple risks if their asylum claims are rejected. 

Finally, automated systems may reinforce bias and discrimination in the migration process. Algorithms are usually trained on historical data that also includes inequalities and discrimination, and might then reproduce these patterns (Gugliotta & Elbi, 2024). For instance, certain nationalities and ethnicities can be associated with increased security concerns. This could lead to targeting and visa refusals against certain groups of migrants, even in the absence of evidence of individual wrongdoing (Endevio, 2025). Additionally, it is important to evaluate each individual’s personal circumstances and  background to finalise the asylum decision. However, automated systems tend to categorise and oversimplify individual cases based on predetermined parameters. And in these cases, crucial details that could have supported asylum claims can be overlooked. 

Policy Recommendations

For AI-assisted systems and automated decision-making to be integrated into migration procedures without undermining human rights and transparency of migration decisions, several policy recommendations should be considered. 

First, the Lithuanian government should be obliged to conduct an assessment of the impact of the introduction of automated decision-making systems into the migration process. Such an analysis would include potential risks related to discrimination, data protection, and compliance with human rights laws. Moreover, these assessments should be made publicly available to improve transparency and public trust. A clear explanation of how automated processes influence migration decisions in each case would also boost transparency. Finally, applicants equipped with the information on how migration decisions are made would be more aware of common mistakes and would be able to better understand the outcome of their applications. 

Second, human oversight must be present and go hand in hand with automated decision-making. While AI-assisted tools can automate tedious work and support administrative processes, they should not replace human judgment. Machines are known for making mistakes and lacking human understanding (Park et al., 2024). And so, final decisions, especially regarding asylum claims, should always be concluded by trained officials who evaluate the full context of each case. 

Third, stronger safeguards for data protection and storage have to be introduced. Biometric and facial data of the migrants and applicants should only be used for clearly defined purposes that are clearly stated to the applicants. In addition to that, the data has to be stored securely and shared externally only with the applicants’ approval. To ensure compliance with privacy regulations, independent authorities should regularly review and assess these practices. 

Finally, the implementation of automated technology cannot proceed without equal investment in the development of human capital. In other words, AI-assisted tools should not replace human workers, and the government should increase the number of trained officials. This would ensure that applicants receive fair and individual assessments of their claims.  

In conclusion, Lithuania’s approach to integrating automated decision-making and AI-assisted tools into its migration governance shows how technology can transform migration management. While important and useful, these technologies also reshape the way migration decisions are made and raise ethical and human rights concerns. Therefore, the policymakers should not debate whether automation should be implemented, but rather how it should be governed. After all, if implemented responsibly, AI tools may help to improve migration management while also supporting fair and transparent migration decision-making. 

References

Douglas, P. (2025, September 19). Understanding Right to Explanation and Automated Decision-Making in Europe’s GDPR and AI Act. Tech Policy Press. https://www.techpolicy.press/understanding-right-to-explanation-and-automated-decisionmaking-in-europes-gdpr-and-ai-act/

Endevio (2025, April 29). The Impact of AI and Automation on Immigration and Border Control [INSIGHTS Blog post]. Endevio.  https://www.endevio.com/en/insights/ai-and-automation-in-immigration

European Commission. (2022, February). The Use of Digitalisation and Artificial Intelligence in Migration Management. https://home-affairs.ec.europa.eu/document/download/72119462-57dd-4544-ab43-99eb2ffc5103_en

European Migration Network. (2025). Migration trends.  https://123.emn.lt/en/migration-trends/

Fathallah, S. (2026). The EU AI Act and the violent logics of border AI. Internet Policy Review. https://policyreview.info/articles/news/violent-logics-border-ai/2062

Gugliotta, L., & Elbi, A. (2024). Will AI “Subtly” Take Over Decision-making in the EU Migration Context? Warnings and Lessons from ETIAS and VIS. European Papers, 9(3), 1018-1047. https://doi.org/10.15166/2499-8249/797

Langeveld, C. (2025, January 20). LITHUANIA – Immigration changes for 2025. Newland Chase.  https://newlandchase.com/lithuania-immigration-changes-for-2025/

MIGRIS. (2020). Migracijos departamentas. Migracijos Departamentas. https://www.migracija.lt/en/faq

Official Statistics Portal. (2026). WAR REFUGEES FROM UKRAINE. https://osp.stat.gov.lt/en_GB/ukraine-dashboards

Park, P. S., Goldstein, S., O’Gara, A., Chen, M., & Hendrycks, D. (2024). AI deception: a Survey of examples, risks, and Potential Solutions. Patterns, 5(5), 1-16. https://doi.org/10.1016/j.patter.2024.100988

Permanent Mission of Lithuania to the United Nations. (2025, September 26). Statement by the Republic of Lithuania at the Security Council High-Level Open Debate on Artificial Intelligence and International Peace and Security. https://un.mfa.lt/en/news/26/statement-by-the-republic-of-lithuania-at-the-security-council-high-level-open-debate-on-artificial-intelligence-and-international-peace-and-security:1433

Prague Process. (2026). Lithuania. Pragueprocess.eu. https://www.pragueprocess.eu/en/countries/878-lithuania

Protecting Rights at Borders. (2023, May 30). PRAB Policy Note III: Walls and High Tech at Europe’s Borders. Danish Refugee Council. https://drc.ngo/media/2cqnt3oq/prab-_-policy-note-_-walls-and-high-tech-at-europe-s-borders.pdf

Radio Liberty. (2022, January 29). Lithuania To Put Cameras Along Entire Border Barrier With Belarus. RadioFreeEurope/RadioLiberty; RFE/RL. https://www.rferl.org/a/lithuania-belarus-border-cameras-migrants/31677158.html

Rees, C. (2022, May 10). Surveillance Drones for Lithuanian State Border Guard Service | UST. Unmanned Systems Technology. https://www.unmannedsystemstechnology.com/2022/05/surveillance-drones-for-lithuanian-state-border-guard-service/

State Data Protection Inspectorate. (2023, November 9). Visa Information System. https://vdai.lrv.lt/en/kiti-eng/darbo-grupes-eng/visa-information-system/

The Seimas Ombudsmen’s Office. (2021, December 7). UPR Pre-session on Lithuania. https://upr-info.org/sites/default/files/documents/2021-12/1._lithuania_the_seimas_ombudsmens_office.pdf

Vavoula, N. (2021). Artificial Intelligence (AI) at Schengen Borders: Automated Processing, Algorithmic Profiling and Facial Recognition in the Era of Techno-Solutionism. European Journal of Migration and Law, 23(4), 457–484. https://doi.org/10.1163/15718166-12340114 

VFS Global. (2026). MIGRIS Update. Vfsglobal.com. https://visa.vfsglobal.com/vnm/en/ltr/news/lithuania-migration-notice

Note. Consensus.app was used to find relevant sources. 

Turkey

Written by Hayrunnisa Çelik

Edited by Davide di Battista

National Migration Context

In recent years, Türkiye has been home to the largest refugee population in Europe, particularly since the Syrian Civil War (Syrians in Turkey: Current Numbers and Return Trend 2025, 2025). It hosts more than 2.3 million registered Syrian refugees, along with close to 222,000 people of concern from other nationalities UNHCR. (n.d.). As such, frequent waves of migration due to its geographical location and being surrounded by conflict zones have put serious administrative pressure on the Turkish migration and asylum institutions (Göç İdaresi Başkanlığı or “GIB”). Following the collapse of the Syrian regime in 2024, the government has put the focus from hosting to active repatriation. 

With this transition came ‘the zero tolerance policy’ for irregular migration. Around 270 Mobile Migration Points (Mobil Göç Noktası) were deployed with biometric systems that allow law enforcement to verify the residency status of individuals in various provinces of Türkiye which eventually led to high deportation rates (T.C. İçişleri Bakanlığı, 2024).

As such, Türkiye presents a distinctly compelling case for the integration of AI and data-driven migration management, particularly in areas of border surveillance, identity verification, and potential refugee status determination (RSD). Several initiatives highlight Türkiye’s technological advancements in migration governance, including advanced border security technologies, ongoing biometric registration systems and evolving digital platforms supporting service provision for foreigners. These areas of technological advancement are also priorities for the EU which offers a shared basis for further collaboration (Yıldız & Açar, 2024). However, these developments remain underexamined, especially in regards to their implications for procedural fairness, human rights, and accountability mechanisms. This article will explore the extent to which AI and automated decision-making systems are used, or are beginning to appear, in Türkiye’s asylum and migration governance. The piece will also argue that although fully automated decision-making may now be non-existent, the growing use of these systems raises significant legal and ethical concerns regardless.

Use of AI and Automated Decision-Making Systems

As one of the key host countries for migrants, Türkiye’s asylum system is governed by the Law on Foreigners and International Protection (LFIP)(n.d.), implemented by GIB which is now operating under the Presidency of Migration Management Akburakcı, N. F. (2025). A defining feature of the system is Türkiye’s geographical limitation to the 1951 Refugee Convention, which restricts full refugee status to individuals from Europe (Introduction to the Asylum Context in Türkiye, 2025). Non-European asylum seekers are processed through alternative statuses such as temporary protection or conditional refugee status.

Although not a leading adopter of AI in asylum procedures, Türkiye has increasingly relied on new technologies in its migration control. This included thermal cameras (located especially in the country’s east borders), and radar surveillance systems. (Esin Küçük, 2025) In Türkiye, the GöçNet system maintains records of 5.5 million foreigners and operates in integration with more than 20 public institutions since 2021 (Esin Küçük, 2025). Meanwhile, the Migration Management Presidency’s ‘YIMER 157 call center’ provides services in multiple languages, which increases the accessibility for the different migrant populations in the country. Additionally, the AI tools within YIMER improve efficiency by categorizing inquiries, directing calls to relevant personnel, and analyzing data patterns to improve decision-making processes (Yıldız & Açar, 2024).

Lastly, Türkiye introduced Mobile Migration Points (MMPs), which were the first of their kind in the world. The MMPs essentially were created to operate in public spaces by enabling authorities to carry out immediate identity and biometric checks through fingerprints of individuals, who were then sent to the deportation centres if their registration could not be identified. 

Human Rights Implications

Even though the LFIP says that each case must be looked at separately, using AI makes it hard to challenge the algorithmic discretion in court. Turkish administrative law requires the judicial review to be available against all actions and decisions of the administration. This leads to show that the administration cannot act arbitrarily and that every action must be based on a legal reason (Türkiye Cumhuriyeti Anayasası, n.d.). But if such systems were to influence or be integrated into decision-making processes, a machine-learning model could affect the decision to deny residency or to deport someone and it would be technically impossible to give a good reason for the GIB. Therefore, the applicant cannot legally challenge it as the system may not clearly explain why it made that decision, which directly threatens the right to a reasoned decision and an effective remedy, because neither the judge nor the applicant can see the logic that goes into putting someone in a certain “risk” group.

Furthermore, the use of highly accurate biometric systems and AI-supported surveillance in Türkiye’s eastern borders also raises serious concerns about the principle of non-refoulement and algorithmic bias. Additionally, the deployment of MMPs raises significant legal and ethical concerns. The broad discretion afforded to law enforcement officers in selecting individuals for checks, often without individualized suspicion, creates a risk of arbitrary enforcement practices. In practice, this may disproportionately affect certain groups, thereby reinforcing patterns of ethnic profiling. Moreover, when individuals are unable to demonstrate valid registration, they may be transferred to removal centres and subjected to deportation procedures, sometimes with limited access to effective legal remedies.

As such the over-reliance on these systems creates risks to fundamental rights of asylum seekers, especially the principle of non-refoulement, which is endangered when automated risk-profiling leads to a fast-tracked “voluntary” return procedure. If an algorithm incorrectly identifies a person as a security threat based on biased data, such as geographic origin, the person may be deported before a human judge can conduct a meaningful review of the specific dangers they would otherwise face in their home country. Furthermore, procedural fairness is even more endangered with the decreasing individualized assessment. Oftentimes, AI systems categorize people into big datasets which fail to take into account the complex and personal situations that lead to their asylum claims. 

Policy Recommendations

In Türkiye, the LFIP emphasizes the right to an individualized assessment. However, the integration of AI into systems like GöçNet introduces a pre-sorting logic. To address the integration of artificial intelligence in Turkish migration governance, several policy measures should be prioritized:

The Presidency of Migration Management must establish a human oversight requirement to ensure that no final decision regarding deportation or residency is made solely by an automated system.

Legal frameworks must also be strengthened by aligning the Turkish Personal Data Protection Law (KVKK) more closely with the GDPR and the EU AI Act. Categorizing AI tools used in border control as “high-risk” would lead to a rigorous oversight and help protect the 5.5 million biometric records currently in use. Furthermore, specialized training should be provided to the judiciary to ensure they can effectively scrutinize algorithmic evidence and safeguard procedural fairness.

Regarding the enforcement, the government should implement non-discriminatory operating protocols for MMPs to prevent ethnic profiling and arbitrary checks. Mandatory Human Rights Impact Assessments must be conducted prior to deploying new surveillance technologies on the eastern borders to ensure the principle of non-refoulement is not compromised. 

The future of migration management in Türkiye will depend on whether these technologies are used to enhance or circumvent the rule of law. Safeguarding sensitive personal data is especially critical for asylum seekers, who often flee persecution. Therefore, any further integration of AI into Türkiye’s migration management should be approached cautiously to avoid undermining refugees’ access to protection.

References

Akburakcı, N. F. (2025). Yapay Zekânın Kullanımından Kaynaklı Sorunların Göç Hukukuna Etkisi. Türkiye Adalet Akademisi Dergisi, 64, 735–768. https://doi.org/10.54049/taad.1809513

Esin Küçük. (2025, April 2). Gatekeeping for Europe: Türkiye’s Tech-Driven Migration Management – Netzwerk Fluchtforschung. Fluchtforschung.net. https://fluchtforschung.net/gatekeeping-for-europe-tuerkiyes-tech-driven-migration-management/

Introduction to the Asylum Context in türkiye. (2025, July 29). Asylum Information Database | European Council on Refugees and Exiles. https://asylumineurope.org/reports/country/turkiye/introduction-asylum-context-turkiye/

Law on Foreigners and International Protection. (n.d.). https://natlex.ilo.org/dyn/natlex2/natlex2/files/download/108031/TUR108031.pdf

Syrians in Turkey: Current Numbers and Return Trend 2025. (2025). Gencdiplomatlar.com. https://www.gencdiplomatlar.com/masalar/syrians-in-turkey-current-numbers-and-return-trend-2025.html

T.C. İçişleri Bakanlığı. (2024). T.C. İçişleri Bakanlığı Göç İdaresi Başkanlığı – Düzensiz Göçmenlerin Tespitini Kolaylaştıran ve Hızlandıran Mobil Göç Noktası Araçlarının Sayısı 162’ye Çıktı. Goc.gov.tr. https://www.goc.gov.tr/duzensiz-gocmenlerin-tespitini-kolaylastiran-ve-hizlandiran-mobil-goc-noktasi-araclarinin-sayisi-162ye-cikti

Türkiye Cumhuriyeti Anayasası, 125.

UNHCR. (n.d.). Refugees and Asylum Seekers in Türkiye | UNHCR Turkey. UNHCR Turkey. https://www.unhcr.org/tr/en/kime-yardim-ediyoruz/refugees-and-asylum-seekers-tuerkiye

Yıldız, A., & Açar, D. A. (2024). Artificial Intelligence and Migration Governance: Navigating Cooperation and Complexity in EU-Turkey Relations. METU Studies in Development. https://doi.org/10.60165/metusd.v51i2.4

France

Written by Alexandra Sima

Edited by Roberta Franzini

National Migration Context

To understand how asylum and migration operate today in France, one must consider a system shaped by movement, bureaucracy, and political scrutiny. In recent years, France has become one of the main destinations for asylum seekers in Europe, placing considerable pressure on institutions responsible for evaluating claims and providing protection (Prague Process, 2025). The French asylum regime therefore operates at the intersection of humanitarian obligations, legal procedures, and the practical demands of managing a high volume of applications. 

In 2024, more than 153,000 asylum applications were registered by the Office français de protection des réfugiés et apatrides (OFPRA), marking the highest number recorded in recent years (Prague Process, 2025). Behind these figures lies a diverse geography of displacement. Applicants arriving in France come from regions shaped by conflict, instability, and economic hardship. Afghanistan remains among the most common countries of origin, alongside applicants from Ukraine, Guinea, the Democratic Republic of Congo, and Côte d’Ivoire (AIDA, 2025). These patterns reflect broader global displacement trends while also illustrating France’s position as a major receiving state within the European asylum landscape.

The institutional architecture responsible for processing asylum claims in France is composed of several key actors. OFPRA serves as the primary authority assessing applications at first instance, conducting interviews and determining whether applicants qualify for refugee status or subsidiary protection (AIDA, 2025). Individuals whose claims are rejected may appeal before the Cour nationale du droit d’asile (CNDA), a specialized court that reviews asylum decisions (AIDA, 2025). Alongside these bodies, border authorities and regional prefectures play a central role in registering applications and managing procedures linked to European regulations (AIDA, 2025).

Despite ongoing reforms, the system faces significant structural pressures. Processing backlogs and shortages in reception facilities continue to challenge administrative capacity, leaving many applicants in prolonged periods of uncertainty. At the same time, France increasingly relies on interconnected European databases–such as Eurodac–to verify identities and coordinate migration management within the broader Common European Asylum System (AIDA, 2025). Together, these dynamics form the contemporary context in which digital tools, biometric databases, and emerging AI-supported systems are increasingly emerging in migration governance.   

Use of AI in Asylum and Migration

In France, migration governance is increasingly shaped by digital infrastructures and algorithmic systems that support the identification, monitoring, and processing of migrants. While fully automated decision-making is not formally used in asylum adjudication, a network of biometric databases, surveillance technologies, and digital administrative platforms now influences how migration is managed across different stages of the process (Moniz, Talwar, Vindrola-Padros, 2023).

At the center of this system are large-scale European databases designed to ensure rapid information exchange between member states (MS). The most significant among them is Eurodac, the biometric database that stores fingerprints of asylum seekers and irregular migrants (Moniz, Talwar, Vindrola-Padros, 2023). When individuals apply for asylum in France, their fingerprints are automatically compared with entries in Eurodac to determine whether another EU country may be responsible for examining their claim under the Dublin Regulation (AIDA, 2025). Planned reforms will further expand this database by adding facial images and additional personal data (Statewatch, 2025).

Automated identity verification technologies also operate at borders and registration points. Biometric scanners compare fingerprints or facial images with data stored in European systems such as Visa Information System, enabling authorities to confirm identities and detect prior visa applications. In parallel, France participates in EU-level surveillance initiatives coordinated by Frontex, which use satellite imagery, vessel-tracking tools, and predictive data analysis to monitor migration routes and anticipate irregular crossings. 

Digital technologies also accompany migrants once they enter the asylum procedure. Administrative platforms are used to manage case files, deliver official documents, and schedule interviews with authorities such as the OFPRA (AIDA, 2025). Biometric checks and database queries help determine whether applicants fall under Dublin procedures, while automated systems may flag potential irregularities in documentation or travel history (AIDA, 2025). Although the final determination of refugee status remains a human decision, these digital tools structure much of the administrative workflow.

This growing reliance on automated systems raises questions about transparency and accountability. Migrants theoretically have the right to understand the reasoning behind administrative decisions, yet algorithmic processes, particularly those related to risk profiling or biometric matching, often remain opaque. Challenging incorrect data or automated alerts can therefore be difficult for individuals navigating the asylum process. Oversight of data protection in this context is primarily carried out by the Commission nationale de l’informatique et des libertés (CNIL), which monitors how public authorities collect and process personal data (EDPB, 2023). As migration governance becomes increasingly digitalized, CNIL’s role highlights the tension between technological efficiency and the protection of fundamental rights within the asylum system.

Human Rights Implications of AI in Migration Governance

As digital systems become increasingly integrated into migration governance in France, their use also raises a number of human rights concerns, particularly in relation to bias, data protection, and procedural fairness. Although these technologies are often introduced to improve administrative efficiency and capacity, their use in asylum contexts–where decisions may determine whether a person is granted protection or returned to danger–requires careful scrutiny. 

One of the most frequently cited concerns is the risk of bias and discrimination. AI systems are not neutral; they rely on datasets that may reflect existing patterns of administrative or societal bias. Risk-profiling mechanisms used in European travel authorization systems, for instance, rely on predefined indicators that may disproportionately flag individuals from certain regions as potential migration-related risks (Manzotti, 2025). Similarly, automated language or dialect analysis tools, occasionally used to verify an applicant’s country of origin, have been criticized for their limited ability to capture the linguistic diversity of many regions (Dumbrava, 2025). Such limitations may lead to incorrect conclusions about an applicant’s identity or credibility.

The use of automated tools also raises concerns regarding the principle of non-refoulement, a cornerstone of international refugee law that prohibits returning individuals to countries where they face persecution or serious harm. If automated systems are used to classify applications rapidly–particularly in accelerated or border procedures–there is a risk that complex personal circumstances may be overlooked (Dumbrava, 2025). Incorrect database matches, flawed risk assessments, or overly broad categorization of “safe” countries could potentially lead to wrongful refusals or transfers without adequate individual examination (AIDA, 2025).

Privacy and data protection represent another major area of concern. The migration management infrastructure increasingly relies on large biometric databases such as Eurodac and other biometric identification systems, creating extensive digital records of individuals who are often already in vulnerable situations. The scale of data sharing between migration, law-enforcement, and border authorities has led to criticism that migration governance risks evolving into a form of continuous surveillance (PICUM, 2024). Oversight bodies such as CNIL play an important role in monitoring how such data is processed and ensuring compliance with European data-protection standards (Schumann, Minano, & Maggiore, 2025).

Finally, automation may affect procedural fairness and access to remedies. Asylum law requires that each claim be assessed individually, yet algorithmic tools function by identifying patterns across large datasets (AIDA, 2025). When administrative decisions are influenced by opaque risk indicators or database flags, applicants may find it difficult to understand the reasoning behind an outcome or to challenge potential errors. For many asylum seekers– who may already face linguistic, legal, or technological barriers–this growing reliance on digital systems risks creating an additional layer of complexity within an already difficult process. 

Policy Recommendations

As digital technologies become more integrated into migration governance in France, their use must remain consistent with fundamental rights. While automated systems can help authorities manage large numbers of asylum applications, clear safeguards are necessary to ensure fairness, transparency, and accountability.

First, strong human oversight should remain central to asylum procedures. Automated tools should support administrative processes but must not replace human decision making. Authorities such as OFPRA and CNDA should retain full responsibility for evaluating applications and reviewing outputs generated by algorithmic systems.

Second, greater transparency is necessary. Applicants should be informed when automated tools influence administrative decisions and should have access to clear explanations about how these systems function. This would strengthen procedural fairness and make it easier to challenge potential errors. 

Finally, regular independent audits of AI systems and stronger data protection safeguards are essential. Oversight by institutions such as the CNIL remains crucial to ensuring that the growing use of digital technologies does not undermine the rights of vulnerable individuals.

References

Asylum Information Database. (2025). Country Report. https://asylumineurope.org/wp-content/uploads/2025/06/AIDA-FR_2024-Update.pdf

Asylum Information Database. (2025). AIDA Country Report on France. https://ecre.org/aida-country-report-on-france-update-on-2024/

Asylum Information Database. (2025). Regular Procedure. France. https://asylumineurope.org/reports/country/france/asylum-procedure/procedures/regular-procedure/

Dumbrava, C. (2025). Artificial intelligence in asylum procedures in the EU. https://www.europarl.europa.eu/RegData/etudes/BRIE/2025/775861/EPRS_BRI(2025)775861_EN.pdf

European Data Protection Board. (2023). Guidelines 05/2022 on the use of facial recognition technology in the area of law enforcement. https://www.edpb.europa.eu/system/files/2023-05/edpb_guidelines_202304_frtlawenforcement_v2_en.pdf

Manzotti, C. (2025). A European language detection software to determine asylum seekers’ country of origin: Questioning the assumptions and implications of the EUAA’s project. https://blogs.sussex.ac.uk/sussex-centre-for-migration-research/2025/01/10/european-language-detection-software/

Moniz, T., Talwar, S., Vindrola-Padros, C. (2023). METHODS USED BY FRANCE, GERMANY, ITALY AND SWEDEN TO STREAMLINE ASYLUM PROCEDURES. RAPID EVIDENCE BRIEF. https://www.cape.ac.uk/wp-content/uploads/2023/07/Dated-POST_RREAL-Asylum-review.pdf

Prague Process. (2025). France.  https://www.pragueprocess.eu/en/countries/854-france?tmpl=component&ml=1

Schumann, H., Minano, L., Maggiore, M. (2025). France spearheaded successful effort to dilute EU AI regulation. https://euobserver.com/3691/france-spearheaded-successful-effort-to-dilute-eu-ai-regulation/

PICUM. (2024). Exclusion by design: Unveiling unequal treatment and racial inequalities in migration policies.  https://picum.org/wp-content/uploads/2024/08/Exclusion-by-design.pdf

Statewatch. (2025). 3. Frontex and interoperable databases. https://www.statewatch.org/frontex-and-interoperable-databases-knowledge-as-power/3-frontex-and-interoperable-databases/

Cyprus

Written by Aagya Tripathee

Edited by Davide di Battista

Abstract

Artificial intelligence and algorithmic tools are increasingly being explored and piloted across EU migration and asylum procedures. These tools are often promoted as a way to improve efficiency, enhance standardisation, and manage the growing migration caseload. However, asylum and migration procedures are highly consequential, having the capacity to severely impact vulnerable people’s lives. Thus, it is essential to consider whether all EU member states are equally equipped to implement such technologies responsibly and whether it is worthwhile to implement them in processes involving sensitive, high-stakes dealings. This article will examine these issues in the Cypriot context, a particularly relevant perspective given Cyprus’s status as a frontline migration country with one of the highest migratory burdens within the bloc. This article will outline the current AI integration and usage within the country’s migratory process, briefly examine the rights implications and conclude with some policy recommendations.

National Migration Context 

Cyprus is located in the eastern Mediterranean, roughly 100km off the coast of Lebanon, the island serves as a geopolitical bridge between the EU and the Middle East and thus acts as a frontline state for migration into the EU. The country is a hotspot entry point for irregular migrants into the EU, both because of its close maritime proximity to the Middle East and because of the presence of the UN buffer zone along the Green Line, whose loosely defined borders facilitate irregular crossings into EU territory. Consequently, Cyprus has some of the highest asylum application rates per capita in the EU alongside one of the largest regular migrant populations within the bloc (European Union Agency for Asylum, 2025; Eurostat, 2024). The result is a disproportionately high migratory burden placed on the country, which, as a smaller state, lacks sufficient institutional capacity to accommodate this. It is therefore unsurprising that Cyprus has adopted a hardline, deterrence-first approach to migration. This stance is evident in the country’s policies, including investments in physical and surveillance barriers, promotion of return initiatives, frequent land and sea pushbacks despite human rights concerns, and suspension of asylum processing for entire national groups (Cyprus Refugee Council, 2025). These are just a few examples of this wider pattern. These attitudes are particularly noteworthy when considering the implementation of automated tools, as AI systems tend to amplify and reinforce existing biases and institutional behaviors.

In terms of AI implementation in asylum and migration processing, Cyprus is more an AI recipient than a driver. The overall extent of AI use and automation remains limited and does not directly influence asylum or migration decision-making. This is largely due to the country’s asylum and migration processing infrastructure, which remains paper-based and dependent on in-person procedures (Migration Department of Cyprus, 2024; UNHCR Cyprus, 2024). Thus, the extent of AI integration within migration systems relies on EU-level systems such as Eurodac and Frontex (Cyprus Refugee Council, 2025). AI is used in EU-wide data integration systems and border management functions, mostly for detection and surveillance rather than direct involvement in case processing and decision-making. In practice, AI integration in Cyprus occurs through three main channels: EU data systems such as EURODAC and SIS, operational support from Frontex through the joint Ledra project, and participation in research initiatives such as REACTION.

EU data integration: EURODAC/SIS

As an EU member state, Cyprus is required to participate in EU-wide data integration systems related to migration and border management. Most notably, it is part of EURODAC and, as of July 2023, has been connected to the Schengen Information System (SIS) despite not being a Schengen member. At this level, AI use primarily exists at the arrival stage, where biometric data such as fingerprints is collected, stored, and analysed. In Cyprus, EURODAC is used to implement the Dublin Regulation by utilizing algorithmic sort and search on biometric data to determine which Member state is responsible for examining each asylum application (European Parliament and Council, 2024). Whereas SIS is used for broader border management and security functions, allowing Cypriot authorities to access alerts related to entry bans, irregular migration, and security risks (eu-LISA, 2026). These systems use the shared Biometric Matching System (sBMS) to apply algorithmic matching to compare biometric data across EU databases (eu-LISA, 2026). Thus, automated processes in these systems are limited to identification and verification rather than autonomous decision-making.

Frontex and the Joint Operation Ledra

Frontex is the EU’s central border management agency, responsible for coordinating and supporting Member States through surveillance, technical and operational support, and by providing risk analysis and intelligence. In Cyprus, it plays a more supportive role within the Ledra Joint Operation initiative, collaborating with the Cypriot police and EUROPOL (Council of Europe, 2023). Specifically, it provides operational support by deploying officers to assist with migratory and asylum burdens at borders and refugee centers (Frontex, 2022). There is no publicly documented deployment of AI systems in Cyprus (Frontex, 2022). The deployment of AI and algorithmic processing systems is limited and mainly associated with the EUROSUR system. Where EUROSUR uses algorithmic processing and AI-supported tools to analyze data from several EU databases and information systems, such as SIS, EURODAC, and satellite-based data from the Copernicus programme, to analyze border crossings, arrivals, weather patterns, and smuggling routes (Council of Europe, 2023; FRONTEX, 2026). Thus, FRONTEX’s use of AI and algorithms in Cyprus is largely confined to the EUROSUR program, which supports Cypriot authorities with intelligence and risk analysis, improving situational awareness and border management decisions (Council of Europe, 2023; BVMN, 2024).

REACTION research incitative

At the regional level, Cyprus is involved in a bilateral research initiative with Greece on the REACTION border surveillance project (KIOS Research and Innovation Center of Excellence, 2024). The project is a collaboration between the Greek and Cypriot governments, national authorities, and academic institutions, funded by the European Commission (Information Technologies Institute, 2024). Specifically, it is an initiative that focuses on developing algorithmic processing software to integrate ground-level surveillance hardware, such as drones and autonomous vehicles, with algorithmic analysis and EU-wide border monitoring infrastructure through EUROSUR (KIOS Research and Innovation Center of Excellence, 2024). REACTION aims to support Greek and Cypriot authorities by providing real-time situational data and emergency alerts (KIOS Research and Innovation Center of Excellence, 2024; Information Technologies Institute, 2024). This technology is intended to operate at the pre-arrival stage, where its role is detection and monitoring, and it is not directly involved in asylum decision-making.

Human Rights Implications

In October 2024, the European Court of Human Rights unanimously ruled that Cyprus had violated the European Convention on Human Rights in the case of M.A. and Z.R. v. Cyprus. This case involved two Syrian nationals whose boat was intercepted at sea by Cypriot coast guard vessels and forcibly returned to Lebanon without their asylum claims being registered, without any individual assessment of the risks they faced, and without any legal recourse to challenge their removal. Ultimately, the Court ruled that by returning the applicants in this manner, Cyprus had breached the prohibition on inhuman and degrading treatment and, crucially, the principle of non-refoulement (M.A. and Z.R. v. Cyprus, 2024).

Cyprus operates within a broader human rights context in which pushbacks are not isolated incidents but part of a recurring pattern (Cyprus Refugee Council, 2025). The country has also been linked to  land pushbacks along the Green Line, thus denying access to asylum procedures and leaving people stranded in the buffer zone. Concerns have also been raised about the “voluntary return” programme, with NGO reports suggesting that some individuals often feel pressured to return (Cyprus Refugee Council, 2025). The country has also shown a pattern of differential treatment based on nationality, including the suspension of asylum applications for entire groups, such as the suspension of Syrian asylum applications in 2024 (Cyprus Refugee Council, 2025). It is within this context of a system that prioritizes deterrence over protection, in which AI systems are being introduced and used.

Surveillance technologies that enhance the speed, range, and precision of detection–such as those used by Frontex and developed under the REACTION project—risk reinforcing deterrence-oriented practices rather than protection. While these systems do not directly make asylum decisions and are therefore difficult to challenge in court (BVMN, 2024), their use in pre-arrival operations can prevent individuals from reaching territory where they can seek asylum, effectively limiting their ability to access the asylum process. Ultimately, stricter deterrence measures can have counterproductive effects. They may push people to take more dangerous routes, delay asylum registration, and increase the administrative burden on authorities managing irregular arrivals. Clear and accessible asylum procedures would likely reduce risks for migrants while also making the system more manageable for the state (UNHCR, 2018).

Many within the EU have publicly criticized the U.S. Immigration and Customs Enforcement (ICE). However, with the development of newer ‘return’ policies and certain activities by Frontex, Europe is not far removed from similar practices (PICUM, 2026; Braude, 2026). Technologies that were once used primarily for search and rescue, like drones, are now increasingly deployed to detect and push back migrants (Guardian, 2019). There has been a well-documented body of evidence highlighting human rights concerns linked to Frontex (Human Rights Watch, 2021). And operationally, there are similarities between the two organizations: both organizations use surveillance technologies, including facial recognition, to track migrants, and both maintain armed personnel (Braude, 2026). While Frontex was once mostly limited to assisting at borders, its role is increasingly expanding toward more direct enforcement, blurring the distinction between the two agencies (Braude, 2026). One of the most widely reported concerns is the alleged systematic concealment of rights violations, including a pattern of failing to prevent, report, or hold perpetrators of pushbacks accountable (Human Rights Watch, 2021). Even with a wide range of internal and external oversight mechanisms in place, Frontex has largely managed to evade compliance with EU and international human rights law.  

The organization does this mainly in two ways, which is highly relevant to the deployment of similar AI surveillance systems like REACTION. Such systems can also evade accountability in much the same way: through blame-shifting and by obscuring transparency.

One of the main reasons why it is so difficult to hold Frontex accountable is its structural diffusion of responsibility. Frontex rarely operates alone, instead conducting joint operations with Member State authorities. This allows the agency to attribute direct responsibility for abuses to national actors while itself claiming the role of just a coordinating body. The absence of a proper framework for joint liability creates this accountability gap (Kusiak, 2025), allowing for serious rights violations to go unpunished. Consequently, blame is often individualised, directed at figures such as Fabrice Leggeri (de la Baume, 2022), rather than addressing broader institutional and technological systems that enable such practices. This concern also applies to other AI surveillance technologies, which rarely operate in isolation and instead function alongside other actors within broader operational networks. Allowing for the evasion of accountability through blameshifting.

​Another way operators of AI and algorithmic systems can evade accountability is by remaining opaque about the operational use of these technologies. It is difficult to hold AI systems accountable when oversight bodies are not even aware of the full extent of their deployment, let alone the criteria guiding their operations—a tactic that Frontex has consistently employed (Human Rights Watch, 2021). Yet, Frontex justifies this secrecy and the withholding of information on the grounds of “public security”, citing the need to maintain the agency’s operational effectiveness (Human Rights Watch, 2024). By framing operations in terms of security,Frontex can evade accountability, even in cases involving human rights abuses. Where redacting information can be enough to dismiss a case, as the Sea‑Watch v. Frontex ruling showed, with the EU General Court backing the agency, thus reinforcing Frontex’s capacity to evade transparency (CJEU, 2024). Furthermore, the usage of “black box” technologies, such as machine learning systems , in which internal processing is opaque, and the reasoning behind outputs is often difficult to interpret, makes transparency challenging (Rinaldi and Teo, 2025).  While the Frontex case isn’t specific to AI, the strategies it employs are instructive.

There is a technological arms race, a rush to develop and deploy AI as quickly as possible. Thus, widening the gap between the pace of AI technology development and the progress of AI policy and regulation. For instance, the EU AI act will only apply to large-scale data systems like Eurosur starting in 2030, even though organisations such as Frontex are already testing and using these systems today (Regulation (EU) 2024/1689, 2024; Frontex, 2026). In a country that has adopted a deterrence first stance towards migration, it is especially important to have effective safeguards with strong legal oversight over AI to ensure that biases do not undermine the obligation to maintain a fair and just asylum and migration system. Legislation for AI and automated systems should make transparency mandatory, clearly define liability and identify who to hold accountable when things go wrong. It should enable joint responsibility among organizations and enforce stronger sanctions. Where oversight bodies should have not just monitoring powers, but also the authority to impose penalties to ensure compliance. Currently, Cyprus lacks a dedicated national AI regulatory framework, with AI governance primarily based on EU legislation, specifically the EU Artificial Intelligence Act (Aristidou & Marcou, 2025; Deputy Ministry of Research, Innovation and Digital Policy, 2025). This is why there is a clear need for national legislation governing the implementation of automated tools and the storage and responsible handling of data. Furthermore, greater funding and promotion should be provided to AI policy and ethics research and initiatives, thus reducing the gap between AI development and AI research. AI regulations should be established prior to the implementation and deployment of AI systems.

Policy Recommendations 

As of now, almost all migratory processing in Cyprus remains paper-based and dependent on in-person procedures (Migration Department of Cyprus, 2024; UNHCR Cyprus, 2024). This creates unnecessary administrative bottlenecks and exhausts the system’s existing capacity. Thus, the country should focus on digitizing and standardizing its migration infrastructure before pursuing more advanced technological solutions.  And despite Cyprus’s efforts to modernise through the Digital Strategy 2020–2025, which commits to the broad digitalisation of public services, the strategy does not explicitly address migration administration. This suggests that targeted investment and political will are needed to modernise the migration processing system.

A central challenge for Cyprus’s migration processing system is the disproportionate burden it bears compared to other EU Member States. As one of the EU’s smallest countries, Cyprus has a small civil service and limited institutional capacity, and it simply lacks the resources to manage such high volumes effectively (Cyprus Refugee Council, 2025). Although the New Pact on Migration and Asylum introduces a mandatory solidarity mechanism, its flexible approach to burden-sharing may be insufficient in redistributing the migration burden. This, combined with the continuation of a responsibility criterion similar to the Dublin Protocol, means that Cyprus will remain a frontline state, bearing the brunt of the migration burden (Gazi, 2021). This is why another recommendation for Cyprus would be to form a coalition with other frontline migration states to demand greater support and more stringent obligations for member states that extend beyond financial contributions. A practical starting point would be to create an EU-level, needs-based civil service pool for migration processing, complementing the existing Frontex Standing Corps. Such a pool would ensure that small states like Cyprus are not left to handle surges alone, with solidarity mechanisms supporting actual migration processing rather than just border control. At the national level, Cyprus could create programmes that incentivize and empower youth to join civil service, helping to grow a skilled workforce more capable of managing migration pressures over the long-term.

As of now, Cyprus is not yet ready to implement AI and algorithmic technologies responsibly, largely due to its underdeveloped migration processing system and human rights background. There is pressure for countries to keep up with AI development and implementation, and while research and innovation are important, Cyprus must carefully consider the risks and potential for misuse. AI could help alleviate some of the burdens Cyprus faces, but the key challenge is ensuring ethical use and strong accountability mechanisms. Cyprus should focus on strengthening its human rights protections before adopting AI technologies it is not yet prepared to manage.

References

Aristidou, C. & Marcou, E., 2025. AI, Machine Learning & Big Data Laws and Regulations 2025 – Cyprus. International Comparative Legal Guides (Global Legal Group). Available at: https://hybridlawtech.com/2020/12/cyprus-fintech-laws-and-regulations-contribution-to-the-iclg-guide/

Border Violence Monitoring Network (2024) BVMN – Border Technologies: Cyprus [PDF]. Available at: https://borderviolence.eu/app/uploads/BVMN-Border-Technologies-Cyprus.pdf

Braude, R. (2026) Europeans Outraged at ICE Should Also Be Resisting Frontex [Online]. Jacobin. Available at: https://jacobin.com/2026/02/europe-frontex-ice-repression-immigration

Council of Europe: European Court of Human Rights (2024) M.A. and Z.R. v. Cyprus (Application no. 39090/20), ECLI:CE:ECHR:2024:1008JUD003909020, 8 October 2024 [Online]. Available at: https://www.refworld.org/jurisprudence/caselaw/echr/2024/148792

Council of Europe (2023) Cyprus – Country factsheet: Smuggling of migrants. Available at: https://rm.coe.int/cyprus-2764-2570-2665-v-1/1680aefb8a

Court of Justice of the European Union (2024) PRESS RELEASE No 70/24: Judgment of the General Court in Case T‑205/22 | Naass and Sea‑Watch v Frontex [Online]. Luxembourg, 24 April 2024. Available at: https://curia.europa.eu/site/upload/docs/application/pdf/2024‑04/cp240070en.pdf

Cyprus Refugee Council / ECRE (2025). AIDA Country Report: Cyprus — 2024 Update. European Council on Refugees and Exiles, April. Available at: asylumineurope.org/wp-content/uploads/2025/04/AIDA-CY_2024Update.pdf

de la Baume, M. (2022) EU border agency chief resigns after critical watchdog probe [Online]. Politico. Available at: https://www.politico.eu/article/eu-border-agency-fabrice-leggeri-resigns-watchdog-probe-findings-released-misconduct/

Deputy Ministry of Research, Innovation and Digital Policy, Republic of Cyprus, 2025. National Strategy for Research and Innovation 2024–2026. Available at: https://www.gov.cy/dmrid/en/documents

eu-LISA (n.d.) Large-scale IT systems. Available at: https://www.eulisa.europa.eu/activities/large-scale-it-systems

European Parliament and Council (2024) Regulation (EU) 2024/1358 of the European Parliament and of the Council of 14 May 2024 on the establishment of “Eurodac” for the comparison of biometric data and related purposes. Official Journal of the European Union, L 1358. Available at: https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A32024R1358

European Union Agency for Asylum (2025). ‘Relative Pressure from Asylum-Related Migration’, in Asylum Report 2025. EUAA, June. Available at: euaa.europa.eu

Eurostat (2024). Migration and Migrant Population Statistics. European Commission. Available at: ec.europa.eu/eurostat/web/products-eurostat-news/w/ddn-20240327-1/

Frontex (2026) Monitoring and risk analysis [Online]. Available at: https://www.frontex.europa.eu/what-we-do/monitoring-and-risk-analysis/monitoring-and-risk-analysis/

Frontex (2022) Public Register of Documents: search results for “Cyprus” [Online]. Available at: https://prd.frontex.europa.eu/

Gazi, A. (2021). ‘New Pact on Migration and Asylum: Supporting or Constraining Rights of Vulnerable Groups?’ European Papers. Available at: europeanpapers.eu/europeanforum/new-pact-migration-asylum-supporting-or-constraining-rights-vulnerable-groups/

Howden, D., Fotiadis, A. and Loewenstein, A. (2019) Once migrants on Mediterranean were saved by naval patrols. Now they have to watch as drones fly over [Online]. The Guardian, 4 August 2019. Available at: https://www.theguardian.com/world/2019/aug/04/drones-replace-patrol-ships-mediterranean-fears-more-migrant-deaths-eu

Human Rights Watch (2021) Frontex failing to protect people at EU borders: Stronger safeguards vital as border agency expands [Online]. Human Rights Watch, 23 June 2021. Available at: https://www.hrw.org/news/2021/06/23/frontex-failing-protect-people-eu-borders

Human Rights Watch (2024) EU misses opportunity on Frontex transparency, accountability [Online]. Human Rights Watch, 24 April 2024. Available at: https://www.hrw.org/news/2024/04/24/eu-misses-opportunity-frontex-transparency-accountability

Information Technologies Institute (2024) REACTION: REal‑time ArtifiCial InTellIgence for BOrders Surveillance via RPAS data aNalytics to support Law Enforcement Agencies [Online]. Available at: https://www.iti.gr/iti/en/project/reaction-en/

KIOS Research and Innovation Centre of Excellence, University of Cyprus (2024). REACTION: Real-Time Artificial Intelligence for Borders Surveillance via RPAS Data Analytics to Support Law Enforcement Agencies. Available at: kios.ucy.ac.cy/projects_kios/reaction-real-time-artificial-intelligence-for-borders-surveillance/

Kusiak, S. (2025) Non‑Contractual Liability: WS and Others as a Blockade to Keeping Frontex Accountable [Online]. Euro Prospects, 8 March 2025. Available at: https://europrospects.eu/non-contractual-liability-ws-and-others-as-a-blockade-to-keeping-frontex-accountable/.

Migration Department of Cyprus (2024). Residence Permits and Immigration Applications. Deputy Ministry of Migration and International Protection. Available at: migration.gov.cy

Platform for International Cooperation on Undocumented Migrants (2026) No to EU law enabling home raids, policing of public services and racial profiling [Online]. Available at: https://picum.org/blog/no-to-eu-law-enabling-home-raids-policing-of-public-services-and-racial-profiling/

Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence and amending Regulations … (Artificial Intelligence Act), Official Journal of the European Union L 1689, 12.7.2024, pp. (Regulation No. 2024/1689) available at: https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32024R1689

Rinaldi, A. and Teo, S. A. (2025) The use of artificial intelligence technologies in border and migration control and the subtle erosion of human rights. International and Comparative Law Quarterly, 74(1), pp. 61–89. doi: 10.1017/S0020589325000090

UNHCR Cyprus (2024). Applying for Asylum: How and Where Can I Apply? United Nations High Commissioner for Refugees. Available at: help.unhcr.org/cyprus/applying-for-asylum/how-and-where-can-i-apply-for-asylum/

UNHCR (2018) Quality asylum procedures, UNHCR Cyprus brochure, May 2018 [Online]. Available at: https://www.unhcr.org/cy/wp-content/uploads/sites/41/2018/05/UNHCR_Brochure_EN.pdf

Italy

Written by Flavia Onwuelo

Edited by Daniele Sorato

National Migration Context

Italy remains the European Union’s primary point of entry along the Central Mediterranean route. In 2024, the country registered approximately 159,000 asylum applications, the highest in a decade, ranking third in the EU after Germany and Spain and representing 16% of all EU applications (AIDA, 2025a). The main nationalities were Bangladesh, Peru, Pakistan, Egypt and Tunisia, i.e. countries with low recognition rates (AIDA, 2025a). At first instance, 78.565 decisions were issued, with an overall recognition rate of 35.9%, a significant fall from 47% in 2022, reflecting both a tightened legal framework and a caseload dominated by low-protection nationalities (ISMU Foundation, 2025).

Since taking office in October 2022, the Meloni government has pursued a restrictive migration policy. The so-called “Cutro Decree” (Law 50/2023) restricted “special protection” status and related reception rights (humanitarian protection having been abolished by the so-called “Salvini Decree” of 2018) (Italian Republic, 2023; Italian Republic, 2024). Decree-Law 145/2024, converted into Law 187/2024, introduced accelerated procedures for irregular entrants, expanded grounds for implicit withdrawal of applications, and transferred jurisdiction over asylum detention validation from specialised courts to courts of appeal, reducing the quality of judicial oversight precisely where it is most needed. (AIDA, 2025b). A state of emergency, originally declared in response to arrivals, was extended repeatedly throughout 2024.

The apex of this restrictive turn was the Italy-Albania Protocol, signed in November 2023 and ratified in February 2024. This agreement established offshore fast-track processing centres in Albania for adult men rescued at sea from designated “safe countries”, positioning Italy as a pioneer in EU-level externalisation and creating the procedural context in which automated and quasi-automated decision-making gained heightened significance (Government of the Italian Republic & Council of Ministers of the Republic of Albania, 2023).

AI and Automated Decision-Making in Asylum and Migration

Italy does not yet operate a nationally integrated AI platform for asylum adjudication. Its engagement with automated tools is primarily EU-driven and occurs at two levels: national deployment of specific AI tools, and participation in EU-wide digital infrastructure. As Ozkul (2023) has documented in a comprehensive mapping of automated systems across European asylum governance, several Member States have begun testing or deploying technologies ranging from dialect recognition to automated evidence-gathering, often without adequate transparency or accountability frameworks.

At the national level, the most documented deployment is S.I.N.D.A.C.A, an AI tool which is being tested by Italian authorities to automatically transcribe interviews. According to its provider, the system processes semantic content including dialects, accents, and spontaneous speech with a claimed accuracy of at least 95% (Dumbrava, 2025). While framed as a transcription aid, automated transcription is not procedurally neutral, as inaccuracies in the rendering of applicants’ words may directly affect credibility assessments.

At the EU level, Italy is deeply embedded in the expanding architecture of interoperable biometric databases. The Recast Eurodac Regulation (2024) requires Italy to register fingerprints and facial images of all arrivals, as part of the EU interoperability framework, which links Eurodac, as a distinct biometric database, with the Schengen Information System (SIS), the Visa Information System (VIS), and the nascent ETIAS (European Parliament & Council of the European Union, 2024). ETIAS includes an algorithmic risk-scoring mechanism capable of generating automated “specific risk indicators”, a form of probabilistic profiling applied to individuals before they reach any human decision-maker (Dumbrava, 2025). Italy also indirectly participates in the EUAA’s ongoing work toward AI-assisted Language Analysis for the Determination of Origin (LADO), which the EPRS Briefing identifies as an emerging deployment across Member States (Dumbrava, 2025).

The Italy-Albania Protocol has introduced another particularly concerning layer to these dynamics: rapid pre-screening conducted aboard Italian naval vessels at sea. Nationality verification and “safe country” classification, both carrying major legal consequences, were reportedly made under shipboard conditions by non-specialist personnel and without adequate legal, medical, or psychological support. These screenings operated as de facto gateway decisions, funnelling individuals into an accelerated procedure strongly oriented towards removal. In October 2024, four of the first sixteen migrants transferred to Albania were returned to Italy after the initial screening failed to identify a minor and individuals with serious health vulnerabilities.  The criteria governing these assessments were never made public (Italian Council for Refugees, 2025; Reuters, 2024; International Rescue Committee, 2025).

Human Rights Implications

The principle of non-refoulement, enshrined in Article 33 of the 1951 Refugee Convention, demands individualised assessment of each applicant’s risk of harm upon return (United Nations, 1951; United Nations High Commissioner for Refugees [UNHCR], 2024). The Albania Protocol’s country-of-origin filtering collapsed this requirement into a group-based presumption. The Council of Europe Parliamentary Assembly has unequivocally warned that AI tools in migration “must not allow for the bypassing of international obligations, in particular under the Refugee Convention”, nor should they be used “to deny safe and effective legal avenues into States’ territory” (Council of Europe Parliamentary Assembly [PACE], 2025). Italy’s shipboard screening, producing near-automatic transfer decisions for nationals of designated countries, crossed this line. The scheme was repeatedly blocked by national courts and the European Court of Justice, which found Italy’s “safe country” designations legally deficient, demonstrating the systemic risk when algorithmic-style logic replace rights-compliant individualised reviews (Reuters, 2024; AIDA, 2025; CJEU, 2024; CJEU, 2025).

The PACE Report explicitly calls for a prohibition on AI tools performing automated credibility assessments or risk profiling based on nationality or ethnicity, finding such approaches scientifically invalid and incompatible with Articles 3 and 14 ECHR (PACE, 2025). The use of S.I.N.D.A.C.A to transcribe and implicitly assess dialect and accent falls within this concern: phonetic analysis offered as objective evidence in a credibility-sensitive procedure, done without independent verification or applicant challenge rights. (Dumbrava, 2025). Molnar and Gill (2018) have documented how these technologies, when deployed against migrants, tend to entrench existing power asymmetries and erode the procedural safeguards that refugee law was designed to guarantee.

Data protection risks are substantial. Italy’s integration into Eurodac and associated systems entails mass collection of biometric data from people in highly vulnerable circumstances. Under Article 9 GDPR, biometric data is a special category requiring heightened protection (European Parliament & Council of the European Union, 2016). The ENNHRI Scoping Paper identifies a lack of transparency in the scope, purpose, and impact of digital migration tools as a key human rights concern, particularly because opacity undermines accountability and access to remedy. At the time of writing, no publicly available fundamental-rights impact assessment appears to be readily identifiable for Italy’s use of S.I.N.D.A.C.A. or for its role in these broader screening infrastructures (European Network of National Human Rights Institutions [ENNHRI], 2024).

Procedural fairness, including the right to an individualised assessment and effective remedy, is systematically undermined when opaque automated tools influence outcomes without disclosure or contestability. The EPRS Briefing identifies inaccuracy and bias as the principal risk categories for AI asylum tools: systems trained on historical data replicate prior patterns of discrimination, while applicants whose cases are processed through such tools have no practical ability to understand or challenge the algorithmic inputs to their decision (Dumbrava, 2025). Italy’s legislative strategy of insulating “safe country” designations from judicial review by elevating them to primary legislation exemplifies how states can deliberately narrow the remedial space that rights-based safeguards require (Italian Republic, 2023; Italian Republic, 2024; AIDA, 2025).

Policy Recommendations

First, Italy should immediately publish fundamental rights impact assessments, including the Council of Europe’s HUDERIA methodology, for all AI and automated tools used in migration and asylum, including S.I.N.D.A.C.A and any screening systems operating at or near the border. The PACE Report recommends such assessments as a prerequisite for deployment, not an optional ex post evaluation (PACE, 2025).

Second, the principle of human oversight must be legally guaranteed in all AI-assisted stages of the asylum procedure. The EU AI Act classifies migration-related AI as high-risk, requiring mandatory risk management, transparency obligations, and meaningful human review, not merely a nominal sign-off (European Parliament & Council of the European Union, 2024; Dumbrava, 2025). Italy should adopt a binding national protocol before the June 2026 entry into force of the EU Screening Regulation, specifying the procedural form and documentation requirements for human review of all algorithmically-informed decisions.

Third, Italy must establish an accessible mechanism for individuals to challenge AI-influenced decisions. As Palmiotto and Ozkul (2024) have argued in the context of strategic litigation against automated migration systems, the barriers to contesting algorithmic decisions are not merely procedural but epistemic: applicants cannot challenge what they cannot see. The ENNHRI Scoping Paper calls for NHRIs to monitor and report on the human rights impacts of technology in migration governance, and the PACE Report recommends expanded legal aid to cover algorithmic disputes (European Parliament & Council of the European Union, 2024; Dumbrava, 2025). Access to legal aid must be guaranteed throughout asylum procedures, including for those held in extraterritorial facilities.

Fourth, at the EU level, the Commission should ensure that the Recast Eurodac Regulation is implemented with robust data governance frameworks, bias audits of algorithmic risk-scoring mechanisms, and clear prohibitions on sharing biometric data with countries of origin where the risk of persecution exists (European Parliament & Council of the European Union, 2024; ENNHRI, 2024). The EU Pact on Migration and Asylum must treat the Screening Toolbox as a procedural aid subject to rights-based oversight, not as a filter for excluding protection claims. Italy, as the primary testbed for these new tools, bears a particular responsibility to model rights-compliant implementation (European Parliament & Council of the European Union, 2024a, 2024b; Dumbrava, 2025).

Conclusion

Italy’s trajectory illustrates a wider European risk: the incremental normalisation of algorithmic decision-making in asylum governance, ahead of the legal frameworks needed to govern it. The four recommendations above represent a necessary baseline for rights-compliant governance under existing obligations under international refugee law, the ECHR, and the EU AI Act. With the Screening Regulation entering into force in June 2026, the window for course correction is narrow.

References

Asylum Information Database. (2025a, July). Country report: Italy – 2024 update. 

European Council on Refugees and Exiles (ECRE). https://asylumineurope.org/reports/country/italy/

Asylum Information Database. (2025b). Overview of main changes – Italy 2024. 

European Council on Refugees and Exiles (ECRE). https://asylumineurope.org/reports/country/italy/overview-main-changes-previous-report-update/

Council of Europe Parliamentary Assembly. (2025, June 24). Artificial intelligence and 

migration: Report (Rapporteur: P. Honkonen; Committee on Migration, Refugees and Displaced Persons). 

https://rm.coe.int/report-artificial-intelligence-and-migration/1680b67b8a

Court of Justice of the European Union. (2024, October 4). CV v Ministerstvo vnitra (Case 

C-406/22). ECLI:EU:C:2024:841. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:62022CJ0406

Court of Justice of the European Union. (2025, August 1). LC v Commissione territoriale di 

Roma; CP v Commissione territoriale di Roma (Joined Cases C-758/24 and C-759/24). ECLI:EU:C:2025:591. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:62024CA0758

Dumbrava, C. (2025, July). Artificial intelligence in asylum procedures in the EU (EPRS Briefing

PE 775.861). European Parliamentary Research Service. https://www.europarl.europa.eu/RegData/etudes/BRIE/2025/775861/EPRS_BRI(2025)775861_EN.pdf

European Network of National Human Rights Institutions. (2024, September). Technologies, 

migration and human rights: The role of European NHRIs – Scoping paper. https://ennhri.org/wp-content/uploads/2024/09/Technologies-migration-and-human-rights-the-role-of-European-NHRIs-an-ENNHRI-scoping-paper.pdf

European Parliament & Council of the European Union. (2016). Regulation (EU) 2016/679 on 

the protection of natural persons with regard to the processing of personal data (General

Data Protection Regulation). Official Journal of the European Union, L 119, 1–88. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32016R0679

European Parliament & Council of the European Union. (2024a). Regulation (EU) 2024/1351 on 

asylum and migration management. Official Journal of the European Union. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32024R1351

European Parliament & Council of the European Union. (2024b). Regulation (EU) 2024/1356 

introducing a screening of third-country nationals at the external borders (Screening Regulation). Official Journal of the European Union. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32024R1356

European Parliament & Council of the European Union. (2024c). Regulation (EU) 2024/1358 

on the establishment of Eurodac (Recast Eurodac Regulation). Official Journal of the European Union. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32024R1358

European Parliament & Council of the European Union. (2024d). Regulation (EU) 2024/1689 

laying down harmonised rules on artificial intelligence (EU AI Act). Official Journal of the European Union. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32024R1689

Government of the Italian Republic & Council of Ministers of the Republic of Albania. (2023, 

November 6). Protocol on strengthening cooperation in migration matters; ratified by Italian Law No. 14/2024. https://www.esteri.it/wp-content/uploads/2023/12/PROTOCOLLO-ITALIA-ALBANIA-in-materia-migratoria.pdf

International Rescue Committee. (2025, February). IRC Italy protection monitoring report: October–December 2024. https://www.rescue.org/eu/report/irc-italy-protection-monitoring-report-october-december-2024

ISMU Foundation. (2025, June). In 2024, asylum applications increased in Italy but recognitions decreased: Only 7.6% granted refugee status [Press release]. https://www.ismu.org/en/in-2024-asylum-applications-increased-in-italy-but-recognitions-decreased-only-7-6-granted-refugee-status-press-release-2025/

Italian Council for Refugees. (2025, March). Beyond the border: The Italy–Albania agreement between propaganda and the suspension of rights. https://cir-rifugiati.org/wp-content/uploads/2025/03/Beyond-the-border.-The-Italy-Albania-Agreement-between-Propaganda-and-the-Suspension-of-Rights.pdf

Italian Republic. (2023). Decree-Law No. 20/2023 (Cutro Decree), converted into Law No. 50 of 5 May 2023. https://www.gazzettaufficiale.it/eli/id/2023/05/05/23G00058/SG

Italian Republic. (2024). Decree-Law No. 145/2024, converted by Law No. 187/2024. https://www.gazzettaufficiale.it/eli/id/2024/10/11/24G00171/sg

Reuters. (2024, October 19). Migrants held in Albania transferred to Italy after court ruling. https://www.reuters.com/world/europe/migrants-held-albania-transferred-italy-after-court-ruling-2024-10-19

United Nations. (1951, July 28). Convention relating to the status of refugees, 189 U.N.T.S. 137.https://www.ohchr.org/en/instruments-mechanisms/instruments/convention-relating-status-refugees

United Nations High Commissioner for Refugees. (2024, September 26). Legal considerations on asylum and non-refoulement in the context of instrumentalization. https://www.refworld.org/policy/legalguidance/unhcr/2024/en/148736

Ozkul, D. (2023). Automating immigration and asylum: The uses of new technologies in migration and asylum governance in Europe. Refugee Studies Centre, University of Oxford. https://www.rsc.ox.ac.uk/publications/automating-immigration-and-asylum-the-uses-of-new-technologies-in-migration-and-asylum-governance-in-europe

Molnar, P., & Gill, L. (2018). Bots at the gate: A human rights analysis of automated decision-making in Canada’s immigration and refugee system. University of Toronto, International Human Rights Program.

https://citizenlab.ca/wp-content/uploads/2018/09/IHRP-Automated-Systems-Report-Web-V2.pdf

Palmiotto, F., & Ozkul, D. (2024). Climbing a wall: Strategic litigation against automated systems in migration and asylum. German Law Journal, 25(6), 935–955. https://doi.org/10.1017/glj.2024.52

Germany

Written by Tamaya Sauerwein

Edited by Marina Pastor Garcia

Abstract

On February 27th, Interior Minister Alexander Dobrindt declared that the governing coalition was “bringing order to migration policy” (Bundesregierung, 2026). A majority in the German Parliament had approved two draft legislation to align national law with the reform of the Common European Asylum System (Tagesschau, 2026a). These entailed stricter asylum rules and the further establishment of so-called “secondary migration centres”, in which austere residency rules apply (Tagesschau, 2026a). Within these centres, rejected asylum seekers see their freedom of movement restricted, which has sparked numerous fundamental rights concerns (Deutsches Institut für Menschenrechte, 2025). Four months after the draft legislation, only three out of sixteen federal states have agreed to set up these centres (Tagesschau, 2026b). 

National Migration Context 

Migration has in fact played a key role in shaping the political landscape in the last decade. In 2015, Chancellor Angela Merkel’s decision to open national borders to refugees and speech “Wir schaffen das” – we can do this – represented a milestone in Germany’s “Willkommenskultur” (Samy, 2026). However, the influx of 890.000 asylum seekers overwhelmed the Federal Office for Migration and Refugees (BAMF) and damaged its institutional reputation (Mrkaja, 2025), while also fuelling a widespread perception that the state lost control over its migration administration. This generally reinforced pressure among policymakers to focus on a more restrictive and tightly regulated migration policy (Angenendt, Koch & Tjaden, 2023). 

The last general elections in 2025 are a prime example of this restrictive migration discourse. The christian-democratic coalition CDU/CSU stated in their election manifesto that Germany needs a “fundamental shift in its migration policy” (CDU, 2025) and that “there is an urgent need for strict limits on migration” (CDU, 2025). And it is precisely under this program, that the CDU/CSU won the elections with 28,6 percent of votes. Ever since, the government’s approach to asylum and migration can be characterized by efforts to regain control over migration. 

However, it is worth noting that these limitation efforts have happened in a context where the number of applications for asylum has decreased. In 2025, the BAMF, which is Germany’s central authority for asylum, migration, and integration, received a total of 113,326 initial asylum applications and 55,307 follow-up applications (BAMF, 2026a). In comparison to 2024, this indicated a decrease of 50.7%, whilst the number of follow-up applications rose by 161.0%, with applicants mostly of Afghan, Syrian or Turkish nationalities (BAMF, 2026a). Despite asylum applications going down, the BAMF still faces many structural challenges, mainly due to its deep bureaucratic nature. Shortage of staff, endless documents and cues: in 2025, the average asylum procedure took 12,2 months (BAMF, 2026a), surpassing the six-month deadline set by the EU law (Regulation 2024/1348). This in turn has favoured the emergence of new technologies to reduce the workload on the migration administration and to speed up administrative procedures. 

The Use of AI in Asylum and Migration 

Artificial intelligence (AI) has been increasingly deployed across migration, asylum and border control, especially in the European context (Angenendt et al, 2023). Its implementation is present at every stage of the administration of migration: from predictive data analysis, through early registration and asylum decisions to final integration mechanism (Molnar, 2025). In Germany, the modernisation agenda for the state and public administration, adopted on October 12025 (Bundesministerium für Digitales und Staatsmodernisierung, 2025), foresees a specific law on the further development of digitalisation in migration administration (Bundestag, 2026). The aim being to “improve data exchange between public authorities in the field of migration management and to speed up administrative procedures” (Bundesrat, 2026). The improvement of efficiency is therefore portrayed as the central argument to justify the usage of AI.

AI and Automated Decision-Making Systems are already employed at various levels, but most frequently in Asylum procedures. The BAMF’s core application, the Migrations-Asyl-Reintegrationssystem (MARiS), a document management system, is for instance used to process all asylum procedures. MARiS handles everything from document management, file distribution, to hearing support (BAMF Digitalisierungsagenda, 2021).

The BAMF also uses the Integrated Identity Management Software (IDM-S) to gather information about the asylum seeker and confirm their identity in case of missing documents (BAMF, 2026b), which constitutes around 65% of all asylum applications (Schattauer, 2026). The use of AI is thus presented as a “system designed to aid authorities in processing asylum applications more efficiently, securely, and accurately” (Van der Kist, 2026).

Automated identification mainly occurs through the processing and comparing of data, which scholars refer to as the “the datafication of migration” (Broeders & Dijstelbloem, 2016, Frowd 2024, Witteborn, 2021 & 2022). Within the EU, fingerprints are matched, compared and stored through the biometric database Eurodac, to ensure that an asylum seeker hasn’t registered in another country and to guarantee the Dublin Regulation, which stipulates that an asylum seeker must apply for asylum in the first European country entered(Eurodac, 2026). In Germany, image biometrics can also serve as means of identification. The IDM-S compares fingerprints and photographs, to ensure that the person does not exist in the system under a different file number or name (BAMF, 2026b). The BAMF also uses automated programs to convert Arabic names into the Latin alphabet, to analyse the name’s distribution and to prevent errors and registrations with different traductions of the name (BAMF, 2026b). In line with these translation mechanisms, machine translation programs are also used during asylum hearings (Molnar, 2025).

Another notable deployment of the IDM-S is the Dialect Identification Assistant System (DIAS) to determine the country of origin of asylum seekers (Van der Kist, 2026). Through an automated language analysis based on speech samples, the DIAS provides different probabilities for possible languages and dialects. It is thereby used to narrow down the region of origin of the asylum seeker, to determine whether they are eligible for asylum and “assess the plausibility of applicants’ statements about their place of origin” (Dumbrava, 2025). 

As a last resort to verify a person’s identity, the IDM-S is also employed to do so-called “phone scrapings” (Mrkaja, 2025). This means evaluating mobile phone data to track all locations and countries in which the phone has been used, and gather all sorts of information about the applicant’s origin (BAMF, 2026b). Among other things, this  includes telephone numbers and their area codes, but also names, addresses, Google Maps searches, call logs, text messages, emails and photos. The IDM-S then creates a report and map tracing back all the locations associated with the analysed data. 

Finally, the BAMF uses the Paas Middleware to provide “a range of services, software components and libraries that significantly simplify the development of cloud-enabled applications in accordance with the BAMF architectural guidelines”(BAMF Digitalisierungsagenda, 2021), in both asylum procedures and integration mechanisms. The Middleware provides services such as logging, file storage and management, archiving and providing data for other systems such as MARiS.

The use of AI and Automated Decision-Making Systems have, as shown above, become entrenched features of migration governance. For this reason, the German government still plans to expand the operational domain of AI in Visa procedures (Bundesregierung, 2025a) and to the Central Register of Foreigners (Bundesregierung, 2025b). 

Human Rights Implications

As Jasper van der Kist notes, AI has changed the way in which governments interact with migrants and refugees, by transforming “the way state administrations manage identities and recognise legal rights and protections” (Van der Kist, 2026). Such transformations raise serious legal and ethical concerns, especially in regards to the protection of migrants’ human rights.

A first concern raised by the implementation of AI in migration governance is the respect of the individual’s right to privacy. Under international human rights law, everyone has the right to respect for their private life, which includes not being subjected to arbitrary interference in personal digital data and correspondence. This fundamental right is enshrined in art. 12 of the Universal Declaration of Human Rights (1948), as well as in art. 7 of the EU Charter of Fundamental Rights (2000). The analysis of personal data in asylum processes in Germany poses serious risks in light of the right to privacy and, in the case of phone scrapping, risks infringing the “Briefgeheimnis” (secrecy of correspondence) inscribed in the German Constitution (art. 10) as well as the Universal Declaration of human rights (art 12). Phone scrapings therefore not only constitute a violation of human rights, but are also simply unconstitutional. The German legal system has also acknowledged these violations within the asylum processes. In June 2021, the Federal Administrative Court declared the repeated use of this procedure to be unlawful and illegal without sufficient consideration of other available findings and documents (Bundesverwaltungsgericht, 2023). 

The lack of transparency in the development and use of new technologies also plays a significant role, as it is usually unclear how migration-related data is processed and used. This is especially problematic with respect to “refugee data”, since the circumstances of forced displacement preclude the possibility of meaningful consent to the collection and use of their data (Molnar, 2025). In case of mistakes made by AI within data analysis or flawed asylum assessments, intransparency makes it more difficult to notice and rectify mistakes, limiting the possibility of recourse. This poses the quintessential question of responsibility and accountability for damages caused by AI. For instance, if automated decisions lead to a disproportionate denial of protection, this would directly contradict the principle of non-refoulement, as set out in the Convention against Torture (art. 3), as well as the individual assessment of asylum applications. And as Molnar notes, it remains unclear who would be made accountable: “the state, a contractor, the software developer, or the person at the immigration authority who uses the application?” (Molnar, 2025).

Finally, the use of AI poses important ethical questions. As  Hannah Davis states, “a dataset is a worldview” (Davis, 2020) – algorithms are never neutral, but rather a reflection of the ideas of the people behind them. AI systems can therefore encode existing human biases. As research shows, risk assessment made by technology is more likely to categorize people of color as higher risk (Mattu et al., 2016). At the same time, facial recognition biometrics tend to misidentify and give inaccurate information while analysing darker skin complexions (IOM, 2021). This can be explained by the fact that AI is trained with certain images, lacking diversity and representation (IOM, 2021). For migration administration, these biases carry enormous consequences, as AI can reinforce structural inequalities, undermine asylum protection, and violate the fundamental right to equality and non-discrimination (Council of Europe, 2025).

Policy Recommendations 

For a state such as Germany, the second-largest destination for international migrants after the United States (IOM, 2021), the potential of AI to solve persisting administrative burdens is significant. In a country where the average asylum-procedure takes longer than a year, the use of AI and Automated Decision-Making Systems is promising, especially in terms of faster and more efficient processing. 

However, it should be noted that the use of AI and automated decision-making systems is transforming migration administration structurally: it influences where priorities are set, and how different migration processes are monitored (Meier, 2026). By doing so, it takes part in determining whose mobility becomes possible (Meier, 2026). As Van der Kist emphasises, “these technologies should therefore not be understood as neutral intermediaries in asylum procedures, but as active mediators that enable and constrain particular ways of doing” (Van der Kist, 2026). Technological instruments are, in this sense, instruments of power (Broeders & Dijstelbloem, 2016), and must be treated as such. Given that AI systems can encode bias and amplify discrimination, and that migrants may find themselves in particularly vulnerable situations as a result of their migration, it is necessary to take a human-rights based approach for migration administration. 

To do so, the first point is to recall that “migration management must balance efficiency gains with protection of human rights at all stages of the migration journey” (Council of Europe, 2025). This implies an impact assessment of human rights before deployment of digital tools, and permanent human oversight during their use. Additionally, big data could and should be used as a tool for the protection of migrants’ human rights. Concrete migration forecasts could, for instance, help prevent deaths at sea and better anticipate situations in which migrants face acute risks. 

Additionally, States need to be transparent about how they use AI in Visa and Asylum procedures, and especially how the data used is processed. Germany should set clear rules regarding transparency and accountability, to guarantee the principle of non-refoulement. Therefore, Germany can largely draw on the EU AI Act, in which migration-related AI systems are classified as high-risk and the need for transparency is clearly and explicitly stipulated (European Parliament et al., 2024b). The BAMF should, for instance, be required to publish regular reports disclosing which AI tools are used and how they process data. This would serve to guarantee applicants the right to contest algorithmic outputs.

Finally, it is crucial that member states of the EU invest in and expand research concerning the benefits and implications of digital migration tools, as it has only been investigated to a limited extent (Meier, 2026). In doing so, Germany and the EU could actively contribute to “help shape fair, transparent, and human rights-compliant standards for the digitalisation of international migration” (Meier, 2026). It is therefore indispensable to keep questioning, in policy and research, how migrants’ lives are being affected by the use of technology, and how it can interfere with the responsibility of a state towards those seeking protection.

References

Angenendt, S., Koch, A., & Tjaden, J. (2023). Die Prognose ungeregelter Wanderungen. Stiftung Wissenschaft und Politik. https://www.swp-berlin.org/publications/products/studien/2023S10_prognose_ungeregelter_wanderungen.pdf 

Beduschi, A. (2018). The big data of international migration: opportunities and challenges for states under international human rights law. Georgetown Journal of International Law, 49(3), 981-1018. 

Biselli, A., & Beckmann, L. (2019). Digitalisierung von Migrationskontrolle. Gesellschaft für Freiheitsrechte. 

Bundesamt für Migration und Flüchtlinge. (2016, November 14). Migrationsbericht 2015. Bundesministerium des Innern. https://www.bamf.de/SharedDocs/Anlagen/DE/Forschung/Migrationsberichte/migrationsbericht-2015.pdf?__blob=publicationFile 

Bundesamt für Migration und Flüchtlinge. (2026a, January 12). Aktuelle Zahlen (12/2025). https://www.bamf.de/SharedDocs/Anlagen/DE/Statistik/AsylinZahlen/aktuelle-zahlen-dezember-2025.html

Bundesamt für Migration und Flüchtlinge. (2026b).  Identitätsmanagement. https://www.BAMF.de/DE/Themen/Sicherheit/Identitaetsmanagement/identitaetsmanagement-node.html 

Bundesamt für Migration und Flüchtlinge Digitalisierungsagenda. (2021). Überblick über die Digitalisierungsinitiativen. https://www.bamf-digitalisierungsagenda.de/wp-content/uploads/2021/03/Digitalisierungsinitiativen-des-BAMF.pdf

Bundesministerium für Digitales und Staatsmodernisierung. (2025, November 5). Bericht der Bundesregierung über Maßnahmen der Bundesregierung für Bürokratierückbau [Cabinet decision]. https://bmds.bund.de/themen/staatsmodernisierung/entlastungskabinett

Bundesrat. (2026, January 30). 1061. Sitzung, TOP 34: Entwurf eines Gesetzes zur Weiterentwicklung der Digitalisierung in der Migrationsverwaltung.  https://www.bundesrat.de/SharedDocs/TO/1061/erl/34.pdf

Bundesregierung. (2015, November 25). Rede von Bundeskanzlerin Dr. Angela Merkel zum Haushaltsgesetz 2016 vor dem Deutschen Bundestag [Bulletin 152-1]. https://www.bundesregierung.de/breg-de/service/newsletter-und-abos/bulletin/rede-von-bundeskanzlerin-dr-angela-merkel-452842

Bundesregierung. (2025a, November 5). Künstliche Intelligenz in Visumverfahren. https://www.bundesregierung.de/breg-de/aktuelles/ki-visumverfahren-2392362

Bundesregierung.  (2025b, December 17).  Im Kabinett: Digitalisierung der Migrationsverwaltung. https://www.bundesregierung.de/breg-de/aktuelles/kabinett-digitalisierung-migrationsverwaltung-2399874 

Bundesregierung. (2026). Bundesinnenminister Dobrindt zu GEAS. https://www.bundesregierung.de/breg-de/service/newsletter-und-abos/bulletin/bmi-geas-anpassung-2409244

Bundestag. (2026). Gesetz zur Weiterentwicklung der Digitalisierung in der Migrationsverwaltung [draft bill]. https://dip.bundestag.de/vorgang/gesetz-zur-weiterentwicklung-der-digitalisierung-in-der-migrationsverwaltung-migrationsverwaltungsdigitalisierungsweiterentwicklungsgesetz/329472

Bundesverwaltungsgericht. (2023, February 16). Pressemitteilung Nr. 13/2023. https://www.bverwg.de/pm/2023/13 

Broeders, D., & Dijstelbloem, H. (2016). The datafication of Mobility and Migration Management: The mediating state and its consequences. In Van der Ploeg & J. Pridmore (Eds.), Digitizing identities: Doing identity in a networked world (pp. 242–260). Routledge. https://doi.org/10.4324/9781315756400-13

CDU. (2025). Wahlprogramm von CDU und CSU. https://www.cdu.de/wahlprogramm-von-cdu-und-csu/ 

Council of Europe. (2025, September 12). Ensuring respect for human rights in the use of AI in migration management. Council of Europe Migration and Refugees. https://www.coe.int/en/web/migration-and-refugees/-/ensuring-respect-for-human-rights-in-the-use-of-ai-in-migration-management

Davis, H. (2020, March 7). A dataset is a worldview. TDS Archive. https://medium.com/data-science/a-dataset-is-a-worldview-5328216dd44d 

Deutsches Institut für Menschenrechte. (2025, December 1st). Menschenrechtsinstitut kritisiert geplante Freiheitsbeschränkungen für Geflüchtete. https://www.institut-fuer-menschenrechte.de/aktuelles/detail/menschenrechtsinstitut-kritisiert-geplante-freiheitsbeschraenkungen-fuer-gefluechtete

Dumbrava, C. (2025). Artificial intelligence in asylum procedures in the EU (EPRS Briefing No. PE 775.861). European Parliamentary Research Service. https://www.europarl.europa.eu/RegData/etudes/BRIE/2025/775861/EPRS_BRI(2025)775861_EN.pdf

EU AI Act. Chapter III: High-risk AI systems. https://artificialintelligenceact.eu/chapter/3/

Eurodac | eu-LISA. (2026). https://www.eulisa.europa.eu/activities/large-scale-it-systems/eurodac

European Parliament, European Council, & European Commission. (2024a). Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act). Official Journal of the European Union. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=OJ:L_202401689

European Parliament, European Council, European Commission. (2024b). Regulation (EU) 2024/1348 of the European Parliament and of the Council of 14 May 2024. Official Journal of the European Union. https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=OJ:L_202401348

European Parliament, European Council, European Commission. (2000, December 18). Charter of fundamental rights of the European Union (2000/C 364/01). Official Journal of the European Communities. https://www.europarl.europa.eu/charter/pdf/text_en.pdf

Equality and Human Rights Commission. (2021, June 22). Article 8: Respect for your private and family life. https://www.equalityhumanrights.com/human-rights/human-rights-act/article-8-respect-your-private-and-family-life

Frowd, P. M. (2025). The ‘Datafication’ of Borders in Global Context: The Role of the International Organization for Migration. Geopolitics, 30(4), 1635–1653. https://doi.org/10.1080/14650045.2024.2318580 

Grundgesetz für die Bundesrepublik Deutschland. (1949, May 23). Art. 10. https://www.gesetze-im-internet.de/gg/art_10.html

International Organization for Migration. (2021). World migration report 2022. United Nations. 

Mattu, J. A., Jeff Larson, Lauren Kirchner, and Surya. (2016, May 23). Machine bias. ProPublica. https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing 

Meier, A. (2026). Migration im Algorithmus: Wie Digitalisierung und Künstliche Intelligenz globale Mobilität verändern. Stiftung Wissenschaft und Politik, German Institute for International and Security Affairs. https://doi.org/10.18449/2026A09

McGregor, L., & Molnar, P. (2023, September). Digital border governance: A human rights based approach. University of Essex & Office of the United Nations High Commissioner for Human Rights (OHCHR). https://www.ohchr.org/sites/default/files/2023-09/Digital-Border-Governance-A-Human-Rights-Based-Approach.pdf

Molnar, P. (2025, September 4). Künstliche intelligenz im migrationsmanagement. Bundeszentrale für politische Bildung. https://www.bpb.de/themen/migration-integration/kurzdossiers/migration-und-sicherheit/570377/kuenstliche-intelligenz-im-migrationsmanagement/ 

Mrkaja, D. (2025, January 16). KI-Einsatz beim BAMF: Teuer, fehleranfällig und selten hilfreich. Der Tagesspiegel. https://background.tagesspiegel.de/digitalisierung-und-ki/briefing/ki-einsatz-beim-bamf-teuer-fehleranfaellig-und-selten-hilfreich

Regulation 2024/1348. Regulation (EU) No 2024/1348 of the European Parliament and of the Council of 14 May 2024 establishing a common procedure for international protection in the Union and repealing Directive 2013/32/EU.  https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A02024R1348-20260227

Samy, Shimaa. (2026, January 29). Fear, power politics and migration: Whatever happened to Germany’s ‘welcome culture’? EUobserver. https://euobserver.com/200162/fear-power-politics-and-migration-whatever-happened-to-germanys-welcome-culture/ 

Schattauer, G. (2026, March 15). Neue Asyl-Zahlen: Fast 75.000 Flüchtlinge kamen ohne Pass. Focus. https://www.focus.de/politik/neue-asyl-zahlen-fast-75-000-fluechtlinge-kamen-ohne-pass_7b111528-51bc-47fe-a61e-ae4a2e02ca1c.html 

Tagesschau. (2017, September 15). Ex-Chef Weise: „Das BAMF steht besser da als 2015“. Die Tagesschau.

https://www.tagesschau.de/inland/interview-weise-bamf-101.html

Tagesschau. (2026a, February 27). Christoph Mestmacher, ARD Berlin, zur Bundestagsabstimmung zur EU-Asylreform. https://www.tagesschau.de/video/video-1559804.html 

Kornmeier,C. (2026b, June 11). Kaum Zuspruch für Dobrindts Migrationszentren-Idee. Die Tagesschau. https://www.tagesschau.de/inland/sekundaermigrationszentren-100.html

United Nations. Universal declaration of human rights. https://www.un.org/en/about-us/universal-declaration-of-human-rights 

United Nations. Convention against torture and other cruel, inhuman or degrading treatment or punishment. https://treaties.un.org/doc/publication/unts/volume%201465/volume-1465-i-24841-english.pdf

Van Der Kist, J. (2026). Algorithmically constructed shibboleths: The technological mediation of automated dialect recognition in asylum procedures. Geopolitics, 31(2), 810–845. https://doi.org/10.1080/14650045.2025.2530417  

Vavoula, N. (2021). Artificial intelligence (AI) at Schengen Borders: Automated processing, algorithmic profiling and facial recognition in the era of techno-solutionism. European Journal of Migration and Law, 23(4), 457–484. https://doi.org/10.1163/15718166-12340114 

Witteborn, S. (2021) Digital placemaking and the datafication of forced migrants. Convergence 27(3). 637‑648. https://doi.org/10.1177/13548565211003876 

Witteborn, S. (2022). Digitalization, Digitization and Datafication: The “Three D” Transformation of Forced Migration Management, Communication, Culture and Critique, 15(2), 157–175.  https://doi.org/10.1093/ccc/tcac007