Author : Pauline Lesch and Pilar De La Torre, The European Forum for Urban Security (Efus)

The European Forum for Urban Security (Efus) is the only European network that offers support, expertise and exchange of best practices in the field of urban security and prevention. The research undertaken by the CCI partners and the solutions developed by local law enforcement agencies inspired Efus to launch a series of web conferences in June 2020 - both as a way to disseminate CCI research and the tools, and as an opportunity to exchange with Efus member cities and regions of our working group on Security and Innovation. 

The web-conferences started a conversation about technologies currently in use or in development in European countries. The goal was to learn about the cities’ point of view: what technologies did they use? What opportunities did they see? What risks and challenges were they facing?

The input from the CCI findings allowed us to go beyond easy narratives that are either exceedingly optimistic or pessimistic - a common pitfall when it comes to polarizing topics such as predictive policing, facial recognition or subjective feelings of insecurity.

Predictive Policing and Facial Recognition

Held on 25 June 2020, the inaugural web conference on predictive policing titled: ‘Technologies based on Artificial Intelligence in Urban Security: Predictive Policing,’ offered participants the opportunity to hear from experts in the field and discuss the effectiveness and impacts of this relatively novel policing approach in EU countries. CCI Partners, Maximilian Querbach, from the State Office for Criminal Investigation of Lower Saxony (Germany) LKA and Dr Oskar Gstrein from the University of Groningen (Netherlands) RUG, spoke about the CCI project's main findings.

At a time when a series of events all over the world triggered an international debate about policing methods, in particular concerning racism and the inappropriate use of force, it is important to understand how ethical concerns are addressed. Dr Gstrein analysed the ethical, legal and social aspects of predictive policing and highlighted issues related to data selection and machine bias, transparency and accountability and the stigmatisation of certain communities and neighbourhoods. Involving these groups in the process and systematically tackling questions related to data selection processes and the fairness of impact assessments could mitigate stigmatisation and increase police accountability.

Alongside predictive policing, facial recognition is one of the most polarizing use cases of artificial intelligence¹. The legal, ethical and social implications are manifold: there is the risk of using historic crime data that might result in automated decisions that reinforce discriminatory bias. The fact that algorithms are trained on a massive amount of data, makes transparency tricky - and understanding where decisions come from too. This in turn impacts the ability to correct erroneous decisions and have recourse to proper judicial review. The use of surveillance cameras can impact the freedom of assembly and of association, as well as the right to non-discrimination. Studies have found that the error rate varies depending on gender and skin colour. In addition, not a lot of research has been done on how facial recognition software works for differently abled people.

What is the impact on the perception of safety?

The CCI project focuses part of its research on the measurement and mitigation of feelings of insecurity. Perceptions of safety and feelings of insecurity were recurring topics in almost every session. Emotional responses range from situational anxiety, over shock and anger, to fear of crime. In order to untangle these responses it is important to understand which factors affect different groups of people. Feelings of insecurity have an impact on individual and collective well-being, on political and economic behaviour.

Considerations on feelings of insecurity lend an interesting angle to the discussion on facial recognition. The presence of security cameras and the knowledge that they are outfitted with facial recognition software can impact inhabitants’ perception of security and even the way they behave in or use public spaces. Some people might feel uncomfortable using public spaces they know to be under surveillance. The societal acceptance of facial recognition technology depends on cultural contexts, differing approaches to privacy and the purpose and context of the use. Whatever the local context and a city’s specificities, local security actors must find the right balance between risks and benefits.

What safeguards to mitigate the risks?

After a first series of discussions on predictive policing and facial recognition, we dove deeper into the topic of artificial intelligence. Our May 5th web conference focused on safeguards to mitigate the risks associated with the technologies. The factsheets on legal, ethical and social implications of predictive policing developed in the framework of the CCI project outlined a number of recommendations, including fostering transparency through clear data management practices and communication with users.

It is important that even those without great technological skills/expertise can comprehend the reasoning behind AI-driven decisions. Representatives from cities and local authorities must, too, be included in development and innovation processes of new AI tools, since they are well-placed to offer insights into the realities of their needs and into best practices for encouraging community engagement.

The panels and leaders responsible for developing and implementing these new AI-based technologies should be representative of the population, ensuring insights in particular from vulnerable groups and minorities - especially at the beginning of the development phase of an algorithm, comprehensive work must be done to ensure diversity and fair representation in data sets and algorithms so as to avoid biases becoming ingrained in these systems.

What technological and social developments foster citizen participation in public policy?

While predictive policing and artificial intelligence are often discussed in great lengths, they are not the only technologies used in the domain of urban security and prevention. Our sessions on civic technology and crime alerting applications moved the discussion to citizen participation and the relevance of feelings of insecurity. The latter topic was further explored in a session on ‘Reducing and preventing the feeling of insecurity at night’ on Wednesday 24 November, organized jointly with the Efus Working Group on Nightlife.

In urban security, civic technology tools can be used to foster an integrated model of security where citizens participate in the co-production of security. The form and scope of interaction between local security authorities and citizens has evolved from one-way and top-down channels of communication to direct exchange and engagement. This entails new consequences, both positive and negative, ranging from rightful denunciation, over risks of increased surveillance, to a disproportionate sense of justice that can degenerate into vigilantism.

Technology as a way to mitigate threats – real and perceived

Mobile crime alerting applications can help victims by providing information on support services, empowering vulnerable groups and potentially encouraging more people to report crime, thus shedding light on previously unreported aspects of crime. Users might feel a sense of security knowing that such a service is at arm’s length. However, while the information gathered through such applications could inform policy responses better tailored to victims’ needs, there are multiple risks. The insights they provide into feelings of insecurity must be considered as an aggregation of subjective perceptions, and as such vulnerable to bias. There is also the risk that members develop a “superhero complex,” particularly in the case of women’s personal safety. In terms of data protection, there is the danger that anonymized data could be compromised through evolving analytic techniques.

Collaboration with local authorities and security and prevention actors can bring greater visibility to existing victim support services. It is also important to acknowledge that statistics based on the data provided by crime alerting applications are not evidence-based but rather add an additional layer to our understanding of the local security situation and the feelings of insecurity of different groups. A user-friendly, multilingual application ensures that the services and information it provides are accessible to a larger group of users. A diverse user-base can ensure that the data is representative of all groups of population.

From research to practice: How can CCI solutions help cities and regions beyond the project partnership?

When we first launched the Security and Innovation web conference series, we relied on the research done by the partners of the CCI project for example via the ‘State of the Art’ reports on predictive policing and measuring feelings on insecurity, the legal, ethical and social implications of policing approaches and the local experiences of law enforcement partners. As the project moved along, so did the focus of the web conferences. We moved from exploration of topics and learning from our member cities to presenting the tools developed by CCI partners and discussing how they could be transferred to other cities and regions.

The first example of this was our March 31st session on Urban Planning, Design and Management of Security in Public Spaces. This web conference was organized jointly with another Efus Project - PACTESUR, which aims to empower cities and local actors in the field of urban security, mainly by enhancing capacities to tackle terrorist threats, but also other risks inherent to public spaces. The aim of this session was to look at the different approaches taken by different cities in different projects.

The terms used to describe multi-stakeholder cooperation and the integration of a security filter during the planning and design phase of an intervention to ensure safety in public spaces vary. The CCI project employs the designation Crime prevention through Urban Planning and Design (CP-UDP). Another commonly used term is Crime prevention through environmental design (CPTED).

When considering the operationalisation of CPTED principles in a public space, it is essential to closely consider the real needs and limits of urban planning. Cities must analyse and improve their levels of knowledge in order to avoid indirect approaches, which can sometimes simply create new problems and further social divisions. Without systematic routines and processes from bottom-up levels, stakeholders may lose control. Safety guidelines must not become a commodified product - they need to be available to all free of charge, so that all communities may benefit. Architects and urban planners must be made aware of their role as public officials, and therefore responsible to the public and for the public, particularly when it comes to safeguarding safety and security.

Looking ahead

The goal of our upcoming web conferences is to create wider awareness, handpick elements that are easily transferable, and develop practice sheets that outline the main steps of how to implement the CCI tools. On Wednesday 9 June 2021, we organised a session on community policing to discuss the insights from the CCI partners and the toolkit they developed. We aim to compare the work done by CCI partners using initiatives of other Efus member cities and regions, in order to identify ways to transfer CCI knowledge.

In addition to the web conferences and the factsheets, Efus is in the process of writing practice sheets about the eight toolkits developed in the CCI project. They outline the main elements of each toolkit and aim to inspire other Efus member cities and regions to take up tools, research methods and insights from the project.

¹. The Efus factsheet on artificial intelligence is available at: https://efus-network.eu/efus/files/2020/11/factsheet_facial-recog_EN.pdf