AI, citizenship and law enforcement to prevent racism, xenophobia and hate speech on the Internet 

On Thursday, June 6 in Madrid, Casa Árabe hosted the presentation of results from the European project REAL UP, for the analysis and creation of alert and response mechanisms against racism and hate crimes. 

The European Project REAL UP Hate speech, racism and xenophobia: alert and response mechanisms, analysis of upstander speech, led by OBERAXE, held its final conference at the Casa Árabe headquarters in Madrid on June 6, expounding its main results, successes and products after 24 months of work.

The project worked to improve the capacities of state authorities to identify, analyze, monitor and evaluate hate speech online, in order to develop and strengthen counter-narrative strategies (upstander speech) against hate speech motivated by racism, xenophobia, anti-Semitism, Islamophobia and anti-Gypsy rhetoric, through debate, creating a map of good practices and effective teaching tools to form the counter-narrative for hate speech online; as well as describing the profiles of perpetrators and victims, a handbook with guidelines to monitor social media safely; the evaluation of the intensity, severity, distribution and potential impact of hate speech, all in order to provide recommendations to deal with upstander speech.

The event brought together professionals in training, data collection, reporting and support for victims of hate crimes and hate speech, representatives of law enforcement, civil society, academia and society in general.

The event began with an institutional welcome speech given by Olivia Orozco, Casa Árabe’s Training and Economics Coordinator; Rafael Pérez Ruiz, the Spanish Interior Ministry’s State Secretary of Security, and Pilar Cancela Rodríguez, State Secretary of Migration.

Pilar Cancela highlighted the project’s contributions to effectively preventing racism, xenophobia and hate speech on the Internet through the development of new tools. This increasing phenomenon in Europe can act as a precursor to violence and threaten democratic values, social stability and peaceful co-existence.

Crime prevention, artificial intelligence and citizen involvement Taking part in the first debate were Tomás Fernández Villazala, the director of Spain’s National Office for Fighting Hate Crimes (ONDOD); Lara Quijano, a professor from the Universidad Autónoma de Madrid, and Rosa Elvira Lillo, a professor from the Universidad Carlos III of Madrid.

This round table highlighted the importance of tackling hate crimes and hate speech from different angles: from the perspective of the Law and crime prevention; and through the social sphere, with the necessary involvement of the people to fight hate speech and discriminatory speech.

The HaterBert algorithm, developed within the project, was also presented. Lara Quijano shared with the attendees the work done to create the algorithm, which makes it possible to tag content published on X (Twitter), categorizing messages into: non-hate, hate, upstander (counternarrative) speech and extreme hate. Quijano stressed the need to train Haterbert and feed it as the language used in social media evolves. The algorithm is seen as a valuable tool that provides support to institutions and authorities so that they can better identify and prevent expressions of hate, while also monitoring users who are considered potential generators of hate. Similarly, Prof. Lillo presented the analysis she has carried out to define the profile of “haters” and victims, based on a series of characteristics, which makes it possible to create predictions about hate speech. An emphasis was placed on the importance of developing new tools to raise awareness and generate knowledge from academia that can be applicable to real life through public administration and civil society.

The use of Artificial Intelligence to create counter-narratives
Taking part in the second round table discussion were Jose Camacho Collados, from the University of Cardiff; Lena de Botton, from the Community of Research on Excellence for All (CREA) at the University of Barcelona; Arturo Lance, from the University of Valencia, and Maite Martín Valdivia, from the University of Jaén.

Professor Martín presented the work carried out with Artificial Intelligence tools, specifically with automatic systems based on machine learning (ML) and Natural Language Processing (NLP), to generate counter-narrative strategies and effective upstander discourse against hate speech.. According to the research, the strategies considered most effective include creating truthful, non-offensive narratives that promote positive discourse as an alternative to attempting to “censor” hate content. Arturo Lance presented the map of tools and good practices that he pointed out should be adapted so that they can be made available and applied in areas such as education.