One capture, many institutions
Eurodac, EES, ETIAS and interoperability link borders, visas, police and asylum.
Paper 37 · Talk · Session 1
Human Rights for All?
AI Facial Recognition at EU Borders and the Limits of European AI Governance
Neil Carelse · Fri 23 Oct, 10:30 · HS102 · LinkedIn (opens in new tab)
Research question
- If human rights are universal, why can't migrants use them when AI facial recognition flags them at Europe's borders?
What must a deployment demonstrate to be rights-compliant?
Contribution
A five-question rights-operational test
Yes or no per deployment site; one 'no' marks the gap to fix before use.
Where formal rights fail in practice
Errors are not evenly distributed
Demographic error differences persist in real border images; nationality, route and language can act as proxies.
Oversight as ritual, remedy too late
Reviewers who can't disagree rubber-stamp; rectification after removal is no safeguard.
Presentation
Slides and talk
Presented on 23 October 2026. The full-screen version includes the live poll results.
Slide viewer
Loading…
Open full-screen talkDownload slides (PDF)
Poster
The poster
Loading…
Download poster (PDF, A0)
Paper
Read and cite
Carelse, N. (2026). Human Rights for All? AI Facial Recognition at EU Borders and the Limits of European AI Governance. In AI Summit Styria 2026 Conference Proceedings. FH JOANNEUM.
Accepted for the AI Summit Styria 2026 Conference Proceedings (FH JOANNEUM). Licensed under CC BY 4.0. DOI to follow.
Download PDF
Literature
Key sources
Molnar, P. (2019). Technology on the margins. Cambridge International Law Journal 8(2).
Migration technology used in opaque, under-regulated "testing grounds".
Migration scholarship
Vavoula, N. (2021). AI at Schengen borders. European Journal of Migration and Law 23(4).
EU border systems future-proofed for facial recognition.
Infrastructure
Grother, P., Ngan, M. & Hanaoka, K. (2019). FRVT Part 3: Demographic Effects. NISTIR 8280.
Error rates vary across demographic groups.
Bias
Buolamwini, J. & Gebru, T. (2018). Gender Shades. PMLR 81.
Gender-classification systems erred far more often for darker-skinned women.
Bias
Atzori, A., Fenu, G. & Marras, M. (2023). Demographic bias in low-resolution deep face recognition in the wild. IEEE JSTSP 17.
Disparities persist in low-resolution images.
Bias
UN CERD (2020). General Recommendation No. 36 on racial profiling.
Profiling can work through proxies.
EU law
Poll
Have your say
- A traveller from outside the EU is flagged by facial recognition at an EU airport. What can she realistically do?
Vote at menti.com, code 3183 9812
Get in touch