Experts told Euronews Next about the impacts of AI systems at the border, saying it deserves more attention from EU lawmakers.
As EU leaders said in Brussels this week that they need to keep exploring new ways to curb irregular migration, experts are calling for more attention to be drawn to how programmes backed by artificial intelligence (AI) and automated surveillance could be used at Europe’s borders.
Experts told Euronews Next that 12 EU states are either testing automated border control systems or AI-powered systems at borders.
Derya Ozkul, an associate professor at the University of Warwick in the United Kingdom, said little is known about how these technologies will impact the rights of migrants.
“Migrants are being kind of used as guinea pigs really in this area,” Ozkul said in an interview.
“All of the glitches with the systems, all of the possible problems and so on are all become faults for migrants to deal with, and there’s not really much accountability”.
What AI tech is already in place at the EU’s borders?
Where AI is already integrated or being tested along some of the EU’s borders is mostly as processing support for asylum seekers, Ozkul said.
One AI tool that Germany’s Federal Office for Migration and Refugees (BAMF) uses is an automated language and dialect recognition software to help officials corroborate a migrant’s country of origin without official documents based on a two minute voice recording.
A statement from BAMF to Euronews Next said the software recognises five major Arabic dialects: Egyptian, Gulf, Iraqi, Levantine, and Maghrebi and could be expanded to more in the future.
The tool is a way of verifying the information that a claimant would provide during their hearing but is only used “in special cases… when appropriate,” the department said.
The department said dialect recognition was used in 43,593 cases in 2023 of the 334,000 applications it received that year.
AI is also used by border officials to help them recognise “security-relevant facts” in asylum hearings, the government said. Decision-makers can then edit, supplement, or discard the information flagged by the AI, so the “human decides, not the AI”.
The AI simplifies the reporting process for decision-makers who say it can be difficult to identify security information during these hearings, BAMF continued.
Ozkul said migrants are often not aware that their asylum cases will be processed using automation or AI.
“[Migrants] usually have no idea how the system is processing their application, and they have no say in the process,” Ozkul said.
“They may not be able to reject the outcome because, again, they don’t have as many rights as citizens would,” she added.
What AI opportunities has the EU identified?
The EU has already invested in some AI-based projects throughout the bloc.
One of them is Centaurus, a project in Greek detention facilities like those along the Evros River that enables CCTV cameras with behavioural analysis algorithms that detect illegal behaviour by groups and uses drones to assess migrant behaviour in the facilities “without human intervention”.
The bloc has also invested in Hyperion, a second system that uses biometric fingerprint data to make it easier to enter and exit detention facilities.
Maria Gavouneli, president of the Greek National Commission for Human Rights, an independent advisory body to the government, called it a “major surveillance exercise that is coordinated [and] financed by the European Union”.
Gavouneli said that most of the surveillance technologies being used at Evros can also be used for national defence, something that will “be intensified” in the coming years.
“The major issue that we are going to be confronted with is questions of transparency, questions of accountability,” Gavouneli said.
The EU has also invested in some AI-supported projects to detect migrant smuggling, like the COMPASS2020 project, which saw patrol vessels autonomously launch underwater drones to boost their surveillance range.
Another, the PROMENADE project, uses AI to improve maritime surveillance and data analysis around 370 kilometres from the coastline.
There’s also a possibility that the bloc expands AI use at borders in the near future, with a few key areas identified in a 2020 white paper.
Automated fingerprint and face recognition, AI-generated emotion detection, migrant risk assessments with algorithms, and AI monitoring of trends are a few key areas of interest, the report found.