Responsible technologies to prevent and curb urban riots

In recent years, the world has seen an increase in riots and violent demonstrations, particularly political and economic protests, against a backdrop of growing inequality and social fractures. To prevent the violence in these movements from marring the often just and necessary struggles, technology can provide a solution by guaranteeing public safety while protecting individual freedom.

By Marie Rollet

Increasing violence in protests

Since 2013, 400 anti-government protests have broken out in more than 132 countries1, with a clear upsurge since 2018. For example, consider the ‘Yellow Vests’ in France in 2018, a pro-democracy movement in Hong Kong in 2019, riots against the high cost of living in Chile the same year, assaults on the capitol in Washington in 2021 and on Brazilian institutions in 2023, and the urban riots in France in 2023. An increasing number of violent episodes claim victims on the side of protesters as well as law enforcement, reflecting the radicalization of the methods used. In South Africa, an unprecedented wave of political violence since the Apartheid was estimated to have claimed up to 330 lives by 2021. Additionally, such movements cause material damage and have lasting effects: for instance, in 2023 in France, approximately 1,000 public buildings were damaged and 6,000 vehicles burnt, worth at least 1 billion euros. Furthermore, they also resulted in business closures and cancellations in the tourism sector, resulting in an estimated revenue loss of almost 2 billion euros. All these have contributed to the deterioration of France’s international image and domestic consumer confidence.2

Whether peaceful demonstrations that get out of hand, or organized, structured movements, these events also reflect an evolution in the forms of protest. ‘Recent years have been marked by the widespread use of hitherto atypical demonstrations, as demonstrated by the Yellow Vests movement, which has resulted in a proliferation of events of a more or less spontaneous nature, with no organizer, no declaration and, therefore, no consultation with the authorities. At the same time, « traditional » demonstrations increasingly seem to be « infiltrated » by violent individuals, whose presence is often completely unrelated to the protest objective, and who undermine the organization of the event, the safety of participants and their right to demonstrate peacefully’3 said the French Members of Parliament in 2021.

Taming social networks

These various movements were amplified by social networks, which have become formidable communication methods to convey demonstrators’ demands and great mobilization tools. ‘The conclusion of the riots in France in the summer of 2023 is that what was said on the networks was decisive for the organization and propagation of the events’, pointed out Guillaume Farde, Professor at Sciences-Po. He adds: ‘Including, in the cases of destruction and looting, for opinion leaders with objectives often far removed from those of the demonstrators, who took local initiatives on the social networks and were followed by small groups of people.’ Consequently, suspension or restriction of access to social media has often been one of the first measures taken to limit the spread of riots and reduce violence, as in Senegal in June 2023, Iran in 2022, repeatedly in Chad since 2016, and Ethiopia in the summer of 2023. In June 2023, Emmanuel Macron, President of France, also mentioned the possibility of cutting off access to social networks ‘when things get out of hand’4, prompting criticism as this proposal echoes the methods of regimes with little respect for public and individual freedoms and because these networks could be a source of early warning of violent movements.

With the necessary technological and legal means to analyse the texts, photos, videos, and other content disseminated on social media, law enforcement agencies can identify opinion leaders and influencers, detect any radicalization of arguments, or monitor the organization of demonstrators to better anticipate gatherings and possible riots. Data-mining tools, which are used by some governments in the fight against terrorism to identify the key words or behavioural patterns associated with radicalization, could also prove useful in preventing urban riots. In France, Suadeo operates the Ministry of the Interior’s big data platform that collects 150 open-source raw data sets, combined with internal sources and feedback from the local prefectures. This enables it to ‘detect emerging signals that give the intuition that an event is about to occur or that it may gain momentum’5, explained Eric Tiquet, Deputy Director of innovation and data at the French Ministry of Interior. Moreover, the analysis of structures and interactions between individuals and social media accounts has benefited from the advances in artificial intelligence (AI) and machine learning, as they ensure more precise and in-depth analyses that could help law enforcement. Bloom’s algorithms, for example, establish technical correlations between content, actors, and communities and track the spread of a conversation.

To ensure a responsible use of these various tools, several countries are working on ‘identifying a core set of security, privacy, and civil rights protections when using social media data and social network analysis in law enforcement’.6

Harnessing the power of AI

The Toolkit for Responsible AI Innovation in Law Enforcement, published by Interpol and UNICRI7 in 2023, is also aimed at ‘support[ing] law enforcement agencies to navigate the complex task of institutionalizing responsible AI innovation and integrating AI into their work (…) based on human rights law, ethics and policing principles’.8 There is a consensus within the law enforcement community that AI is now a useful and efficient tool. As early as 2018, participants at the first Global Meeting on the Opportunities and Risks of Artificial Intelligence and Robotics for Law Enforcement recognized that AI can help enhance how efficiently they can acquire, analyse and act upon information and ultimately improve crime prevention and control.9 Predictive analysis, an interesting use of AI, can easily be applied to riot prevention. Current applications and explorations include “evaluation of risk factors and predict potential of online sexual abuse, monitoring of drivers for radicalization in specific communities and predicting individuals at risk of radicalization, and enhancing CCTV capabilities by enabling the prediction of movement or behaviour of individuals10 through non-visual data, etc. Similarly, machine vision applications are quite popular among the law enforcement community, particularly facial detection and validation tools and surveillance instruments that can monitor sounds, movements, and occupancy as well as detect and report anomalies.11

Since a growing number of cities are introducing smart CCTVs designed to make public spaces safer, urban riots can be prevented using them. The Israeli company Viisight develops an advanced AI-based behavioural recognition system based on CCTV cameras that helps identify crowds of specified sizes and understand their behaviour in a given context and environment. It thus makes it possible to understand how quickly an event is growing, detect suspicious behaviours likely to turn into violent actions, predict possible future behaviours, and then determine the need for crowd management. In compliance with civil rights and data protection regulations, the solution in such cases is strictly limited to detecting behaviours rather than identifying the individuals involved. AI-powered CCTVs can also help detect individuals carrying weapons or different kinds of equipment for instance.

These systems are more problematic in countries where there are no regulations to guarantee individual freedoms and privacy. In Russia, for instance, the Emergency Situation Ministry is believed to be developing an AI software to detect and prevent, if not avoid, the escalation of riots and protests based on the analyses of media reports and data from social networks, CCTVs, and other sources. This is part of the country’s ‘Safe City’ project, which has become one of the most pervasive surveillance systems in the world and is now said to have become a tool to prevent people from protesting.12

On the contrary, when used in a fair, accountable, transparent, and explainable manner, AI-powered tools can become an efficient means to protect genuine protesters against possible troublemakers. Some companies have succeeded in offering solutions that combine operational benefits with ethical commitment. ‘Beyond the legal framework, we are committed to integrating ethics by design into our products, to put technology at the service of people. To this end, we have set up since 2021 an ethics committee comprising members of our teams and external contributors, and published an ethics charter reflecting this commitment’ reveals François Mattens, VP Public Affairs and Strategic Partnership at XXII: beneficence, non-maleficence, autonomy, justice, explicability, and transparency govern XXII’s products. This shows that it is possible to combine ethics and AI, security, and the protection of freedom and privacy.

Extinguishing anger

However sophisticated and effective these technological solutions may be, they are no substitute for political, social, and economic measures that can be a lasting response to the protesters’ demands. These incidents also reflect, in their specific causes and demands and their manifestations, a clear dissatisfaction with political stagnation and growing social inequalities. On October 26, 2023, French Prime Minister Elisabeth Borne unveiled her ‘anti-riot plan’ designed to guarantee citizens’ safety, support families, and strengthen national cohesion. The agenda involves no technology but the establishment of a ‘Republican Action Force’, a tougher penal response, and the creation of new employment integration establishments and the professionalisation of social mediation.

1 Global Protest Tracker – Carnegie Endowment for International Peace

2 Émeutes et flambée de la dette publique : l’été risque d’être chaud et la rentrée caniculaire–

3 Rapport d’enquête n°3786 – 15e législature – Assemblée nationale (

4 Émeutes après la mort de Nahel : Macron blâme… les jeux vidéo (

5 Suadeo implémente sa plateforme Self Data Services au Ministère de l’Intérieur – YouTube

6 Using Social Media and Social Network Analysis in Law Enforcement: Creating a Research Agenda, Including Business Cases, Protections, and Technology Needs | RAND

7 United Nations Interregional Crime and Justice Research Institute


9UNICRI :: United Nations Interregional Crime and Justice Research Institute

10 Ibid

11 Ibid

12 How facial recognition is helping Putin curb dissent (