METICOS aims to provide a monitoring and prediction platform for border surveillance technologies’ social impact, “an up-to-date acceptance classification scheme as well as a societal and ethical impact dashboard of border control technologies, to empower three major sub-divisions of the wider border control sector: travelers, border control authorities and service providers.”
“Performance and credibility expectations of smart border technologies from travelers, border management and law enforcement agencies” would be explored in detail.
“Efficiency” is presented as a key objective. The METICOS consortium had identified “the need” for “no gate crossing point solutions” in order “to allow for the seamless crossing of borders and security checks.”
All of this should happen “without the risk of invading people’s privacy, and gain the societal and political acceptance of these technologies, taking into account the human factor, the social and societal aspects, providing a secure environment for travel and for host countries.”
METICOS highlights the necessity to develop “a platform that integrates information systems and networks of data sources in order to validate the efficiency and users acceptance of border control technologies.”
Privacy, ethical, and societal perceptions are therefore considered metrics and KPIs (key performance indicators) for the proposed platform, while efficiency remains key: “METICOS knowledge base will create a Technology and Societal Acceptance Toolkit with multiple utilizations in favour of law enforcement and border authorities for a better and more efficient Entry/Exit System” (infographic).
The primary focus on efficiency also appears in the project description in deliverable D2.2, which reads:
“METICOS aims to develop a platform that integrates information systems and networks of data sources in order to validate the efficiency and user’s acceptance of border control technologies i.e. fingerprint & iris scanner. The primary purpose of these technologies is to improve the effectiveness and efficiency of border controls, provide more accurate data on entry and exit systems of the EU, speed up board crossings, to examine the efficiency of the new border technologies, the societal and ethical impact on the travellers, including non-EU citizens, the satisfaction and acceptance of both border authorities/managers and travellers, and ultimately to lead in “no gate crossing point solutions” that allow for the seamless crossing of boarders and security checks for the vast majority of travellers.”
The project objectives include:
1) “Indicators for the social and cultural impact of no gate solutions,” “a multi-agent simulation tool to model, analyse and predict the behavioural patterns of the travelers,” “methods for managing the flows of migrants in EU area;”
2) “enhancing security of Schengen borders” (“by providing the means to monitor durations of stay and the identification of persons found to have exceeded their authorised stay” “and by providing the possibility to reconstruct relative historical data”);
3) the “validation of user’s acceptance and satisfaction” (“in order to understand and predict acceptance, as a critical step towards ensuring adoption”);
4) the “effectiveness and efficiency of smart borders” (automated preparation of the border check);
5) “commercialization and exploitation.”
METICOS’ Reporting page on Cordis claims that “METICOS integrates a series of state-of-the-art technological achievements”, going “beyond the state of the art with respect to modelling methodologies and supporting tools.”
It also claims to be able to harmonize and intelligently analyze data from various sources through “a portfolio of Machine Learning, Deep Learning and Statistical algorithms as well as Data Analytics techniques, to uncover patterns regarding acceptance of Smart Border Control Technologies for both border control staff and travelers and to expose the results of the analysis to end-users through an interactive dashboard platform.”
Similarly to many other EU-funded projects in the field of border security, seamless communications and data exchanges are pursued, for example through a “secure middleware” module, while “a planning and decision-support tool for border-control authorities (…) will monitor and predict acceptance (by travellers and staff), secure positive societal impact and maximize border control process efficiency.”
When evaluating the development of such technologies, human rights are contentiously considered in a trade-off with the needs for security and efficiency: “A balance between traveller’s satisfaction, acceptability and human rights and privacy levels and efficiency of deployed security and screening measures will be considered.”
Maximization is another key: “METICOS maximises the returns from EC’s investment in science and technology. METICOS addresses and provide solutions for border control and management planning and optimization thanks to the technology acceptance and social sensing models. Finally, METICOS allows creating new job opportunities that will have an economic impact on the citizens’ life”.
Data protection and freedom of speech could also become an issue as, according to deliverable D3.2, public tweets (and possibly data originating from other social media sources) as well as online reviews on Blogs/Websites regarding acceptance of border control technologies by travelers are collected. The goal is to derive the sentiments that people have towards border control.”
D3.2 details some of the modules developed for the METICOS solution: “The Embedded Technology Acceptance Prediction (ETAP) module quantifies social acceptance of a specific border control technology solution by predicting the likelihood of acceptance/non-acceptance by a person belonging to one of the two targeted groups (travellers, border authorities).”
A “social sensing tool” is then described as “a technology acceptance prediction system for border control technologies. The social sensing tool will analyze and estimate the activity and interactions between migrants/travelers, border staff and border control technologies for technology acceptance prediction.”
An example of user profile data from Twitter (“users who posted tweets related to the border control technologies”), including demographic statistics and user perceptions — which also means their sentiment and emotion — is provided in a YouTube video. The system’s architecture is detailed in D4.4, confirming an interest in emotion recognition technology by following a prior analysis of iBorderCtrl outputs by METICOS staff.
METICOS is the initiator and lead of the so-called “BES (Border External Security) Cluster” of Horizon 2020 and Horizon Europe projects, which include D4FLY, ITFLOWS, ROBORDER, TRESPASS, NESTOR, BORDERUAS, MIRROR, EFFECTOR, PROMENADE, ISOLA, PERSONA, ODYSSEUS, ALIGNER, PERCEPTIONS, CRiTERIA, and more recent additions such as I-SEAMORE, MELCHIOR, ODYSSEUS, iMARS, and FLEXI-cross.
On the project site, we learn that the BES Cluster “consists of projects collaborating in order to support each other, identify solutions to upcoming challenges, secure effective dissemination and valuable exploitation potentials and at the same time generate knowledge that along with the developed solutions will change the current state in the areas and fields that the projects are working upon”.
“Close Collaboration with the technologies and platforms of other BES topics” is one of METICOS’ aims. In fact, “synergies” have been developed with ROBORDER, CRiTERIA, NESTOR, and many other EU-funded projects of interest.
It remains unclear what these “synergies” are about. We do however know from deliverable D3.7 that, for example, “feedback has been requested and gathered from METICOS consortium partners but also from experts from projects in the BES cluster” on the identification and mitigation of AI bias.
“METICOS will be demonstrated and validated under real operating conditions”, while working “closely with border control research initiatives”.
It ran “four pilots across Europe in Romania, Estonia, Cyprus, Greece, and Austria. The METICOS pilots aim to provide valuable outcomes, promoting the use of modern ICT technologies (“no gate solutions”) in the future border management ecosystems.”
Use cases include application trials
1) on border line validation for border authorities, to “evaluate the no gate crossing processes (applied to air/ land/ sea borders); evaluate the acceptance of border & custom authorities of no gate crossing technology; evaluate the social acceptance of no gate crossing technology from travellers point of view; demonstrate that could provide enough data flow and services/tools for analysis;” and
2) on the “validation of EU BCP risk assessment platforms,” to “evaluate the provided enhanced situational awareness in terms of the timely and proper identification of potentially dangerous people and goods, and of prevention smuggling and human trafficking; evaluate the risk-management coordination and cooperation between border control (passport/persons), customs (baggage/goods) and security in transport (pre-boarding security checks on persons and baggage); evaluate the improvement of proposed solutions for remote detection of abnormal behaviours; evaluate the improvement and the acceptance of border automated screening systems.”
Test sites include the airports of Athens, Larnaca, and Moravita — and a “pre-pilot” in Tallinn.
The “METICOS system showcased both in real life and in artificial environment. For the real-life evaluation, an eye-tracking system was used to detect the glances and focus of the wearer when approaching an Automated Border Control (ABC) gate. This approach aimed to understand passenger orientation and recognition of displayed information at the gates. For the border guards, it was investigated how often they look at the METICOS dashboard and which specific areas of the dashboard they focused on. For the artificial environment, METICOS used a Virtual Reality (VR) system where the pilot area was recreated with an ABC gate based on face recognition. This mock-up study should allow a standardised setup to check the orientation and impression of a potential ABC technology. The METICOS system was integrated as a website with the relevant dashboard. Each of the pilots took about 3-4 hours at each pilot site and with 18-22 subjects. The data is currently evaluated, and the results will support the final development of the METICOS system in terms of usability and visualisation of data. Finally, there will be a pilot in Vienna, Austria, to receive results from the final prototype. This will be done in August 2023.”
D3.7 also mentions the “Potential use of Facial Emotion Recognition (FER) during pilots in order to detect the travellers’ acceptance” — all the while noting that this might have a further impact on human rights. However: “After ethical examination, this option was discarded.”
In the Reporting on Cordis, the summary and the overall project objectives tellingly lament that the “major challenge for METICOS is the collection of data for measuring the acceptance of such technologies, directly from the Smart Border Control sites, but without the use of sensors (e.g. cameras) due to the regulatory constraints. Moreover, the evaluation of modern biometric matching technologies through the use of VIS, SIS, EES, etc. or the estimation of Border Authorities’ or travelers’ trust in these procedures through emotion or image analysis is impossible due to ethical, regulatory and security barriers.”
An “Ethics Advisory Board” (EAB) is reported to have been installed. DPOs from the different partners have been identified and have collaborated with the EAB. EAB’s task is to ensure compliance of every ethical requirement and need according to the procedures and criteria defined.
Deliverable D3.2 provides a first “Human Rights Impact Assessment” for METICOS in order to “identify, evaluate and address the adverse effects on human rights emerging from the METICOS project, and in turn, prevent the METICOS components from breaching fundamental rights and/or negatively impacting socio-cultural factors” (note how “adverse effects” are not mentioned as “potential risks” but facts; elsewhere in the document, however, it says “potential impacts on human rights,” ed.).
It is recognized that the project’s very existence may represent a threat to human rights: “Potential impacts on human rights may derive from the overall objective of the METICOS project but also from the big data tools and methods being used.” Consequently, the “scope of this HRIA should take into account potential impacts on human rights deriving from the border control technologies envisaged to be monitored by the METICOS platform.”
The fundamental idea behind the “No-Gate Border Crossing Point Solution” is described as such: The solution is understood to be “a method which enables travellers to cross borders without being slowed down by physical checks, based on automated risk-assessment and/or biometric verification processes.” Does this mean that the risk assessment is — or ideally should become — fully automated?
A list of currently deployed “no gate” solutions includes BorderXpress, BKC-Kordon, the Thales Gemalto EU EES Border Management System, PARAFE, easyPASS, and RAPID.
D3.2 lists potential risks for human rights derived from similar EU-funded projects: SMILE, PERSONA, TRESPASS, MIRROR, PROTECT.
Crucially, D3.2 and D3.7 reveal that METICOS wanted to use, and prioritize, facial emotion recognition technology. During early discussions within the METICOS consortium, “it was considered to use cameras with facial emotion recognition as a primary data source for the METICOS platform. The aim was to use such technology to analyze facial expressions of travellers (and participants during the pilots) in order to measure their acceptance of border control technologies. After ethical examination, this option was discarded” for reasons which included non-proportionality, lack of data accuracy, lack of fairness, lack of transparency and control, risks related to the processing of special categories of personal data (e.g., deriving political opinions and health data), and risks related to profiling and automated decision-making.
However, in a YouTube video posted after this decision (in August 2023), a researcher claims: “We conduct social sensing activities for travelers and border staff to analyze perceptions, such as emotions and sentiment of user profiles, such as age, gender, nationality in a privacy-preserving manner” (1:07).
D3.7 provides a final version for HRIA, Emotion recognition is both rejected and wished upon:
“Within METICOS, in order to reflect the human rights’ concerns deriving from the use of facial emotion recognition, it is recommended to:
• Not to use to use cameras with facial emotion recognition as a primary data source for the METICOS platform (what about non-primary data sources?, ed.). At this stage of the project, it is considered that the use of cameras for behavior & emotion detection could potentially lead to a “function creep”. This means that such data might have been used by border control authorities beyond the limits of the research purposes in order to support measures or decisions taken with regard to travelers (profiling/risk management/etc.).
As a reminder, the purpose of the METICOS platform is to explore both travelers’ and border control staff ‘s attitudes about border control technologies in order to measure technology acceptance and efficiency. The proposed platform will provide aggregated metrics and KPIs to authorities and decision makers, based on a number of independent variables: performance expectancy, effort expectancy, facilitating conditions, physical privacy, accuracy, information privacy, as well as ethical and societal perceptions. The use of facial emotion recognition cameras could have led to the processing of data beyond the scope of the project by providing end-users with data permitting individual (and real-time) decisions that could provoke negative consequences for travelers (for example, further security checks without legal justification in the Border Schengen Code).
• As an alternative to the use of FER, to analyze further the possibility and relevance of collecting data via “traditional” cameras (or physical presence of researchers) on the basis of the informed consent of pilot participants. Such an additional source of data might provide interesting information about the concrete interaction of travellers with border control technologies.”
Furthermore, D3.7 says that the EC’s ethical review required to “submit as a deliverable a report providing an ethical analysis of the AI systems developed or incorporated into the project technologies and the corresponding mitigation actions to address the ethics issues” — and specifies that this is done even though “the METICOS platform does not qualify as “high-risk AI system” in the sense of article 6(2) and Annex III of the proposal AI Act.”
