Only one proposed border security research project did not meet the EU’s ethical requirements and was rejected, AlgorithmWatch found. What was so unique about this surveillance system?
The EU’s research and innovation in border security often involves advanced surveillance systems, aimed at providing full “situational awareness” along its borders. Through the Horizon 2020 (H2020) and Horizon Europe (HE) programs, the European Commission has funded many projects that leverage AI, unmanned vehicles, and data integration from diverse sources to achieve this goal.
To get EU funding, all projects must pass a strict ethics review. This process only once led to a project being rejected on ethical grounds, according to an investigation by AlgorithmWatch. Inquiries to the Research Executive Agency (REA) − the EU body overseeing research projects − revealed that a rejection occurred in the H2020 program. As of the end of 2023, all Horizon Europe projects were given a free pass.
Rejected was Proposal Number 88335, “Enhanced Maritime Situational Awareness through Collection and Integration of Multiple data sources” (called EMERALD). It was designed to augment maritime surveillance using a combination of multiple data sources. In this, it did not stand out: other projects had similar goals, such as integrating data from various sources for improved situational awareness, and were funded nonetheless.
So why the different treatment? What would the EMERALD output have looked like, and why the exclusion from EU funding?
By information requests, AlgorithmWatch obtained 12 partially redacted documents about the rejected proposal. We were denied access to 3 more, on grounds of public security, privacy, and commercial interests. In addition, only 3 out of the 90 pages of the project proposal itself were actually disclosed, leaving only the abstract visible. Even the acronym EMERALD was consistently redacted in the documents we obtained, and only revealed in an accidental mention in a REA email.
One of the three pages we received from EMERALD's Grant Agreement, itself heavily redacted (including the project's acronym)
We did however unearth some useful information. For example, we found that EMERALD aimed for an 18-month duration and that it requested nearly 6 million euros. While the identities of the consortium members who were responsible for the decision were redacted, we learned that they included “industry, academia, and governmental end-users.” The project was to deal with “classified information” and conduct large-scale tests in 4 real-world scenarios: drug trafficking, maritime rescue, terrorist operations, and “surveillance in a complex natural environment.”
EMERALD strongly resembles other EU-funded projects, both in substance and in style. Similarly to H2020 and HE projects such as MARISA, EFFECTOR, and I-SEAMORE, EMERALD was, thanks to “advanced algorithms” and “virtual reality,” supposed to be capable of intelligently integrating data from “small maritime patrol vessels, unmanned aerial vehicles and unmanned underwater vehicles, both with very long endurance and an extended range of operation,” and other sources including satellites and sensors.
Both the goal — “complete” maritime surveillance — and the jargon — “cost-effective,” “lower cost equipment,” “effectively execute missions” — reflect the same solutionist and ultimately anti-democratic fantasy of total control.
EMERALD was rejected due to its failure to focus exclusively on civilian applications—an essential criterion for eligibility under H2020 and HE. According to the disclosed documents, “proposal [X] clearly claims that the enhancement of civil-military interoperability and the improvement of the competitiveness of Europe’s defense industry are major objectives.” The REA’s review emphasized that these references to military aspects conflicted with the requirement for a civilian focus.
The project also raised “a large number of complex ethics issues,” including research involving vulnerable participants (such as refugees and asylum seekers), data privacy, potential risks to humans, and the environment and even a lack of detail on “the measures to take to mitigate these risks.” The EMERALD proposal was even deemed to have “the potential for malevolent/criminal/terrorist abuse.”
Part of the REA's rejection letter, detailing the "complex ethics issues" that the proposal could not properly address.
And yet, the rejection letter said that these “other ethics issues […] were not part of the grounds for rejection,” which was entirely for the lack of an exclusive focus on civilian applications.
This substantial objection was challenged by arguing that the term “defence” in the proposal was “unfortunately chosen” and referring to civil aspects, not to the military. The reviewers did not buy this explanation.
AlgorithmWatch, however, found other instances in which military bodies were involved in H2020- and HE-funded projects, even though the projects were required to focus on civilian applications.
For example, an official document we obtained through access to information requests shows that in the ROBORDER project, which was funded, the risk of a misuse by criminals or terrorists was admitted, too. For this reason, it would be “of the utmost importance that a robust system is in place to ensure the work of the project is not exploited for subversive means.” Other potential ROBORDER outputs were earmarked for the “defence sector” and “military units.”
And yet, ROBORDER received funding. Which leads to the question: How binding are the criteria for deciding if proposals are accepted or rejected?
In fact, concerns about such dual-use applications are quite common and regularly raised. Conflating civilian and military applications is contributing to further facilitating the militarization of EU borders. Human mobility is being transformed into a national security matter and handled invariably as a threat. Among EU decision-makers, the conviction is widely held that the best solutions in this area are necessarily technological and do not need any substantial public scrutiny.
A key point from EMERALD’s Ethics Summary Report makes you sit up and think: “The duality of the military performing civilian activities by its very nature demonstrates the impossibility of an exclusive focus that is fully motivated by and limited to civil applications.” What ultimately means that the very reason for which EMERALD was rejected is applicable to the lot of other projects whose outcomes the military might happily exploit.
Want to learn more about EU-funded experiments at our borders? Read our long-read article:
Read more on our policy & advocacy work on ADM and People on the Move.
