We Need to Start Talking About Artificial Intelligence and Statelessness

September 2025

In this blog, Dr Jason Tucker, Researcher at the Institute for Futures Studies and Adjunct Associate Professor at the AI Policy Lab, Umeå University, Sweden, argues that statelessness studies has largely been silent on the impacts of AI on stateless persons – leading to a blind spot that needs to be addressed.  

Artificial intelligence (AI) is fundamentally reshaping how states operate. AI is altering core state functions, such as identity verification, surveillance, border control and public service delivery. These shifts are not only reshaping the nature and practice of citizenship, but also statelessness. While the societal transformations resulting from increased adoption of AI is one of the most pressing global policy issues, stateless studies and policy debates remain conspicuously silent on the subject. In a recent piece published in the Statelessness & Citizenship Review, I argue that neglecting the intersection of AI and statelessness creates a critical blind spot, where harm can deepen, invisibly and irreversibly. As such, we need to talk about how AI is impacting stateless persons, and the concept of statelessness, and we need to talk about it now.

Statelessness in the Age of AI

States around the world are investing heavily to integrate AI into their bureaucracies. From digital ID systems, biometric databases, and automated decision-making tools in civil registries and public services, AI is becoming central to how many states function. This is nothing new, states have been automating various aspects of their bureaucracies for decades. The difference now is the scale and speed of this adoption, the shift to AI systems that are not transparent, and the reduction in human oversight. This has already, and will continue to, fundamentally alter the concept of citizenship, how citizens and non-citizens interact with states and vice versa.

This impacts stateless persons – significantly. Let’s take one recent example. In July 2025 the UK Government announced that it was going to trial an AI based facial recognition tool to help verify the age of migrants. Facial recognition technology is notoriously inaccurate,  and well known for  replicating racism and bias. For example, research has shown that this technology disproportionately misclassifies individuals with darker skin tones as older than they are.

The UK will use an off-the-shelf AI system - designed to assess the age of shoppers, and likely plagued with racial bias - to set a person on a path which determines if they are able to benefit from extra safeguards to protect stateless children. For example, if a child is incorrectly categorised as an adult they will not be able to benefit from lower fees for citizenship application. This is especially concerning considering the growing evidence on the links between statelessness and poverty. This decision will be made with little to no transparency in how the decision was made and virtually no way to appeal the outcome. Such decisions further risk compounding discrimination faced by stateless persons, most of whom belong to minority groups and who face intersectional and multifaceted discrimination. This is just one example from the broad range of AI applications across the UK migration system that impacts stateless persons.

Data Legacies of Statelessness in AI

Even if statelessness is resolved through the acquisition of citizenship, its impacts linger as data holds a legacy of discrimination. Stateless individuals leave behind a data trail collected during their time without nationality. These datasets, embedded in global data markets, are not deleted or updated once someone gains citizenship. Instead, they are sold, blended, repurposed, and used to train AI systems. This phenomenon, and its impacts, can be understood in relation to broader issues of data bias. For instance, Criado-Pérez has highlighted how medicine has historically been based on male bodies, resulting in a lack of reliable data on women’s health. This “male bias” in data has had serious implications for the development of equitable medical care. In a similar way, stateless communities suffer from what I termed “citizenship-bias”: a systemic lack of quality data about their lives and needs due to their historical exclusion from state systems.

As a result, even after gaining citizenship, formerly stateless people may continue to be excluded from services or opportunities governed by data-driven systems. This is because, similar to what we are seeing in healthcare, AI tools trained on historical data may replicate patterns of discrimination or simply fail to recognise these individuals as eligible users due to gaps in the data. Unless AI systems are explicitly trained to correct for these biases, or data is chosen with great care, they risk creating and compounding legacies of statelessness which will remain embedded in the digital infrastructures that increasingly shape social and political life.

The Path Forward

These are but two of many areas of concern regarding AI and statelessness which I discuss in my recent article. The use of AI across state functions has and will continue to radically alter the identification and protection of those who are affected by statelessness. It is becoming apparent that we are living through a global paradigm shift resulting from the scale of AI adoption by states, and the impacts of this shift on stateless persons cannot be ignored.  As such, we need to discuss how to actively shape and not just be shaped by these changes. This is because the use of, and negative outcomes from, AI is not inevitable. We choose how to develop and where to deploy the technology. On the one hand it can be used to support research and practice related to stateless persons. On the other it can be deployed to oppress and marginalise various groups in society.

For those working in stateless studies the challenge now is for us to map the use of AI systems in society and identify where they impact stateless persons. To do so we need to form interdisciplinary groups to be able to research and audit these systems and work towards mitigating bias, discrimination or harm. We also need to highlight areas where the introduction of AI is too risky, or ill-suited to solving the problem at hand. The deployment of facial recognition technology to assess the age of migrants in the UK exemplifies the risks of applying biased AI tools to high-stakes, rights-based decisions.

That said, AI is not a singular entity but rather a toolbox of continually evolving technologies. Different AI tools are suited to different tasks, each carrying specific capabilities, limitations, and trade-offs that must be carefully assessed. When appropriately selected, responsibly developed, and rigorously monitored, AI can serve as an asset in addressing the complex challenges associated with statelessness.

A Call for International Coordination

While my recent article highlights the critical role that scholars and practitioners in statelessness studies must play, their efforts alone amount to little more than a drop in the ocean. Without international coordination and strategic guidance, these initiatives will remain fragmented and insufficient. UNHCR must assume a leadership role - both technically and normatively - in line with its mandate on statelessness. This includes developing robust policy frameworks that explicitly address the intersection of AI technologies and statelessness; producing best-practice guidelines on algorithmic auditing and stateless sensitive model development; investing in capacity building; and fostering interdisciplinary and cross-sectoral partnerships capable of shaping human rights-based and evidence-based policy on AI in the field.

We are facing a rare and critical window of opportunity to bring statelessness onto the global AI governance agenda. In August 2025, the UN General Assembly approved the establishment of an Independent International Scientific Panel on AI and launched the Global Dialogue on AI Governance. UNHCR, alongside civil society actors and researchers, must act swiftly and collaboratively to ensure that the intersection of AI and statelessness is meaningfully addressed within these emerging frameworks. This is a golden opportunity, but one that demands immediate, coordinated, and sustained engagement. Failure to do so risks rendering stateless persons even more invisible, more marginalised, and more irreversibly embedded within the opaque digital architectures of the age of AI.

_{Image by Steve Johnson on Unsplash}