For Migrants, Artificial Intelligence Becomes a Roadblock to Europe

It was the twilight hours of a midsummer morning near the northern French city of Calais when a group of men sat quietly at a bus stop. A small light aircraft buzzed overhead. Nearby, police on high-speed buggies zoomed up and down the quiet beaches while a drone had been spotted above.

This is just a taste of the millions of dollars worth of technology pumped into trying to prevent these men, who had all failed that night to get onto a smugglers dinghy, from leaving northern France and claiming asylum in the UK. Faheem*, a 42-year-old economics graduate from Sudan living displaced in Calais, said he would continue to try to cross despite having failed multiple times. Many of those failed crossings, he believed, came at the hands of a drone he thought alerted police to his movements.

Utopia 56, an organization that supports displaced people in northern France, said that many of those on the move had resorted to hiding from drones in old World War II shelters along some of the beaches in order to avoid being spotted and turned back by the police. 

French and British authorities appear to increasingly rely on drones to spot boats. In the UK, drones have been used to support prosecutions of people who are criminalized for driving the dinghies into British territory. A Home Office procurement notice from September 2024 suggested drones will play a key role in criminalization moving forward. The notice stated that the drones should be capable of collecting “high-quality footage of criminality to support potential prosecutions.” 

A Deadly Crossing

The stretch of water between France and the UK is one of the busiest shipping lanes in the world, and despite money poured in to prevent migrant departures, 2024 was the deadliest year on record for this migration route since 2014: 82 people died or went missing, according to the International Organization for Migration. 

In a quiet park in the Calais city center in July 2024, Inkstick attended a memorial service for 21-year-old Dina Al Shammari, who was crushed to death in one of the overcrowded dinghies that same month. Her name was added to a long scroll of deaths by NGO workers, which stretched several meters when laid out on the grass.

Al Shammari’s death came just a few months before the British government announced further funding for border security, noting technology would play a key role in the effort to “break” smuggling networks. The UK is already providing around £476 million (around $584.5 million) under a three-year deal made with France in March 2023 under the previous British government to assist the “fight against human trafficking, people smuggling, and illegal migration.” 

“Lethally Dangerous”

In some of the French border towns in the north, a series of infrared cameras in the region, reportedly bought with British funds to the tune of €1.3 million (about $1.33 million), are an example of how the funds are being used. French contributions remain undisclosed but are reportedly “substantial and continuing.”  

A Home Office spokesperson described the English Channel crossings as “lethally dangerous and totally unnecessary.” The spokesperson added, “The UK’s investment in border security and technology enables tens of thousands of lives to be saved each year that would otherwise be in danger of being lost as a result of these criminally facilitated crossings.”

As public opinion across Europe hardens against immigration, the UK is not alone in turning to technology. The months-long joint investigation by Inkstick, Solomon, El País, WoZ, and Tagesspiegel, illustrates how Europe’s dependence on tech-heavy solutions justifies long-held fears about privacy and rights violations.

AI and Border Surveillance

Tech firms are increasingly collaborating with European governments on migration-related projects with Anduril Industries being just one of them. Anduril, an American defense technology firm headed up by Donald Trump supporter Palmer Luckey, has provided a series of AI-powered Maritime Sentry Towers stationed along the south coast of England which are similar to a series of towers positioned along the US-Mexico land border. 

According to Anduril, the towers leverage “the latest in artificial intelligence, machine learning, and computer vision to identify border crossings” and can detect vessels 20 kilometers (12.4 miles) offshore.   

The British Home Office has declined to answer how many towers it has in use from Anduril. However, it said that data collected from the towers is not used in criminal proceedings. The Home Office  also did not respond directly to a query about how the data was being used and if it was being used to further train Anduril’s own technology. Freedom of Information requests to the Home Office suggest that no data protection impact assessments or human rights impact assessments had been done on the Anduril towers. When reached for comment, the Home Office would neither confirm nor deny this.

“Technological Testing Ground”

The panopticon of technology can be seen from the southern British coast to the southern tip of Europe in Greece, where a Greek official told the investigation that British authorities had been impressed with the border technology they’d seen in operation in Greece.

A police report seen by this investigation provides another example of border tech in action. It describes the apprehension of a group of 12 migrants who were spotted about to cross the Evros River on the Greek-Turkish border in 2022. Upon arrival, they were apprehended by a unit of the police who had been alerted by the automated system — they had been watched before even setting foot on the inflatable boat, thanks to an automated border surveillance system Greece has installed, which now covers much of the border and can penetrate 15 kilometers (9.3 miles) into Turkish territory. Three were later charged as smugglers. 

At a gathering in Warsaw in late 2024, law enforcement representatives from across the EU lavished praise on Greece’s success in controlling migration at the Evros land border, highlighting the effective use of “technical barriers,” according to a source. These barriers include a 5-meter-tall steel fence, covering a large part of the 192-kilometer (119-mile) land border, complemented by a sophisticated array of technologies such as AI-equipped drones, omnipresent cameras, and rapid-response teams. 

The watchtowers and surveillance antennas that litter this border were recently mapped for the first time by the research group Forensis, using satellite imagery, public tenders, and open source material. The Border Violence Monitoring Network watchdog called Evros a “technological testing ground” for Europe in a report published in October 2024.

Real-World Implications

These tools not only appear to make it more challenging for migrants to cross the river but also allegedly prevent many from even reaching it. Camera feeds are relayed to monitoring hubs near border cities, where officers behind screens watch much of the border and the area beyond. 

When surveillance drones or cameras pick up activity, an alarm is triggered and Turkish officials are informed to assist in preventing crossings with the Greek side providing coordinates based on shared maps according to descriptions by security officials. The Turkish side often responds, but even when no reply comes, Greek patrols step in. 

These tools not only appear to make it more challenging for migrants to cross the river but also allegedly prevent many from even reaching it.

This kind of collaboration was confirmed to the investigation by both Greek and Turkish officials. Turkey had apprehended more than 225,000 migrants by the end of 2024, according to official data, which does not break down the methods of apprehension. 

A Greek official with direct knowledge of the border systems being used explained that they act as sleepless watchers, which improves border personnel efficiency, while their AI capabilities are able to “interpret what cameras record, for example identifying weapons.” Ultimately, the same official said, the decision on how to act remains with officers.

The real-world implication of what this sleepless surveillance system could assist with has been extensively documented over the years: Greek authorities have routinely faced allegations of violence and asylum seeker pushbacks, charges the government in Athens has consistently denied. 

Reached for comment, the Greek migration ministry said that it does “not provide answers, especially in this phase of heightened immigration, for sensitive operational issues that concern the security of the country.”

Phones Confiscated

This investigation also found that there has been frequent removal of asylum seekers’ phones on arrival to Greece, which was confirmed as a common practice by Greek security sources. 

Three young Syrian asylum seekers interviewed on the Greek island of Samos, a common arrival point, said that their phones, as well as those of everyone they knew, had been seized by the authorities and returned without explanation. They were never told why and did not sign any consent forms. This procedure violates Greek law and the General Data Protection Regulation. “They kept mine for a week after asking me to unlock it,” one of the asylum seekers said. “Others were asked to write down their passwords alongside the phone model.” 

Security sources familiar with these practices confirmed that confiscations occur regularly without court orders or proper documentation, as required by law. Data extracted from phones can be used in criminal proceedings, as well as risk assessments and reports by security agencies, such as the Greek police, Frontex, and Interpol.

Three such reports based on extracted phone data were reviewed. One report, authored by Frontex, included information and pictures from social media and texts from messaging apps found on migrants’ phones to map smuggling networks. A second report, this one compiled by the Greek police, included geolocation data, messages with facilitators and photos of tickets, and itineraries. The third report said migrants had offered their passwords “voluntarily.”

“The Hellenic Coast Guard follows strict legal procedures regarding the confiscation of mobile phones, ensuring that actions are carried out in full compliance with national law,” the Greek coast guard said by email, adding that the “rights of migrants are respected” throughout the process. 

“Most Invasive”

Greek law enforcement officials admitted even private photos and other data are sometimes accessed. Even when they had been deleted from their original device, they could still be accessed from the cloud storage system. One official said law enforcement was not interested in looking into people’s private photos in this process — nonetheless, this casualness regarding migrants’ privacy appears at odds with the preoccupation of Europeans with protecting their own personal data. 

In Germany, data extraction from phones and the use of contentious AI technology are part of the standard asylum process. Greek sources who spoke to the investigation on condition of anonymity described the Federal Office for Migration and Refugees (BAMF), the German agency responsible for processing asylum claims, as one of the most technologically advanced asylum agencies in Europe, including in the use of AI systems. On its own website, BAMF claims it is “a pioneer in the digitalization of the asylum process,” boasting its “innovative technologies.” 

The German agency employs several such systems: automated dialect recognition, AI-assisted reviews of asylum applications, and extensive data collection from mobile phones. BAMF said that the systems are used “exclusively to support the asylum procedure,” while decisions on asylum applications “will continue to be made exclusively” by human specialists.

Despite assurances, these practices have been criticized for their impact on the fundamental rights and privacy of asylum seekers, low efficiency, and high cost. Stephan Scheel, a professor of Political Sociology at Leuphana University of Lüneburg in Germany, has researched the identification technologies used by BAMF. Scheel says mobile phone data extraction is “the most invasive” of the systems used by the agency.

Legal Concerns

While asylum seekers can theoretically refuse the extraction, this could result in the termination of their asylum process according to documents asylum seekers are required to sign This trove of their personal data — including calls, photos, passwords, and more — is analyzed by automated systems and often used during asylum interviews to evaluate the legitimacy of claims and trustworthiness of asylum seekers. The legality of this practice has been a subject of contention since German law permits data extraction only when asylum seekers lack valid identity documents. 

The system did not provide usable results in 73% of cases in the first half of 2023. Despite being declared unlawful by a federal court in early 2023 unless less intrusive methods are used to identify the person, BAMF continues to sidestep the ruling.

The Dialect Recognition Assistant System (DIAS), intended to verify if asylum seekers originate from the regions they claim, has garnered interest from other countries, including Greece. “This is a tool we would like to have,” said a Greek official with direct knowledge of the country’s asylum system. The official described this technology as “very useful.” 

Hanne Beirens, Director of Migration Policy Institute Europe, said that there should be caution around the use of technology and AI in migration management, which she said “raises critical issues of human rights, cost efficiency, and policymaker expertise.” 

Beirens explained, “Technology is increasingly deployed for border control and deterrence, and investment is focused on deterrence of persons entering the territory, which of course often comes at a high human cost while overlooking the broader socio-economic drivers of migration,” she said. “Effective solutions require transparent reporting, independent monitoring, and a shift toward cost-effective, root-cause-focused investments like regional partnerships.” 

Undeterred

The long arms of technology, however, seem to do little to dissuade people from pursuing their journey through Europe. Officials as well as NGO workers who spoke to the investigation all cast doubt on the current power of AI to do much in the near future to prevent arrivals. A group of 12 men Inkstick spoke with in northern France said that they intended to continue their journeys, citing their family in the UK, the fact that they already knew English, or a history of British colonialism that had, they argued, pushed them to Britain’s doorstep. 

One man, a Kurdish Turk, spoke of how he had been bitten by police dogs in Bulgaria, abused by police in Serbia, and pursued by drones in Hungary. He said he hoped that things would be better in the UK. 

Another young Sudanese man, Mustafa* looked out of his tent, intrigued by the talk of drones and people’s difficult journeys. “Is there any other way?” he asked of the dangerous crossing to the UK. 

He was met by silence. 

*All migrants’ names have been changed. 

Note: This investigation was supported by grants from the Investigative Journalism for Europe Fund, Journalism Fund Europe, and Netzwerk Recherche.