Spain’s New Approach to Monitoring and Preventing Online Hate
The Digital Dynamics of Online Hate
Online hatred can take many forms. Messages, videos or even emojis that dehumanise individuals or that indirectly incite violence against people or communities are, unfortunately, on the rise. This reality is particularly concerning when hate is directed at people based on their identity or perceived identity, such as their gender, ethnicity or religion. International human rights law explicitly guards against discrimination on these grounds, reflecting their historical association with persecution and exclusion. The protection of these groups reflects the recognition that such forms of hatred can have a profound impact on individuals’ dignity and the safety of vulnerable communities. This is why, for instance, a social media content that disseminates ideas of racial superiority raises particular human rights concerns[1]. In Spain, recent trends indicate that people from North Africa and people of African descent, as well as communities like Roma and Jewish people, are among those more commonly targeted by online harmful discourses (some of which may rise to the level of hate speech prohibited under Spanish law or the EU’s Digital Services Act)[2].
Beyond harming people, these discourses also affect the quality of democracy. Hate can have a chilling effect on political participation, as fear may push vulnerable communities to refrain from engaging in digital spaces, where much of the political and social debate takes place.
Digital spaces are distinctive in how amplification occurs. The factors that explain the rapid dissemination of hate content online cannot be separated from real-world contexts, where social frustrations and existing tensions often play an important role. At the same time, digital spaces have peculiarities that deserve attention. In particular, digital platforms are governed by specific mechanisms that influence how information spreads. For example, algorithmic amplification prioritises content that generates higher levels of engagement, potentially contributing to the formation of “information bubbles” in which users are repeatedly exposed to similar viewpoints and narratives. Over time, this limited and self-reinforcing information environment may intensify pre-existing beliefs and facilitate gradual processes of radicalisation.
In addition, repeated exposure has a normalising effect. When discriminatory language or hostile narratives become frequent in users’ feeds, they may gradually be perceived as less shocking or more socially acceptable. Even individuals who do not initially agree with such messages can become desensitised over time, making it easier for harmful stereotypes and exclusionary narratives to circulate without being questioned.
Platforms are owned by private companies whose business models often rely on maximising user interaction. As hate content frequently provokes strong emotional reactions, it generates the engagement that feeds this business model, particularly for platforms that have shown a clear disregard for the social damage that their model causes. In the town of Torre Pacheco, where far-right extremist groups organised “hunts” targeting migrants in July 2025, content removal by platforms was reportedly minimal despite the fact that these messages, combined with disinformation surrounding an alleged assault of an elderly man by migrant youths, were likely to contribute to the violence that followed. According to the Spanish Observatory on Racism and Xenophobia (OBERAXE), more than 33,000 racist hate messages were recorded during the week of the events[3] [4].
Another distinctive dynamic in digital platforms is that the actors behind content creation and narrative seeding can be easily obscured. As a result, it can be difficult for citizens to distinguish between spontaneous content and coordinated influence campaigns, or between authentic participation and coordinated inauthentic behaviour. This opacity increases the potential for manipulation as citizens may be unaware of the campaigns designed to influence them. Moreover, users are exposed daily to harmful content without necessarily recognising it as hate or fully understanding its impact.
Spain’s Emerging Framework for Monitoring Online Hate
A recent article in El País[5] highlights Spain’s efforts to address the phenomenon of online hate in the country – a challenge that is part of a broader global trend. The article argues that these efforts place Spain at the centre of emerging initiatives aimed at understanding and addressing the spread of hateful content online. Indeed, in recent years, Spain has developed monitoring and reporting tools to support the institutional identification and analysis of hate speech online.
Among these milestones, the FARO System (Filtering and Analysis of Hate on Social Media) is a new tool used by OBERAXE since 2025 that allows for the real-time identification and analysis of hate speech content reflecting racism, xenophobia, Islamophobia, antisemitism or anti-Roma bias[6]. The system was developed following an agreement in 2024 between the Spanish Ministry of Inclusion, Social Security and Migration and LALIGA[7], allowing for the incorporation of advanced artificial intelligence trained through LALIGA’s Monitor for the Observation of Hate in Sport (MOOD).
FARO has recently been complemented by a new tool: Huella del Odio y Polarización (HODIO, “Footprint of Hatred and Polarisation”). According to institutional sources, HODIO will produce a regular ranking of platforms comparing users’ exposure to hostile content on each platform[8]. In other words, the tool will support, in practice, platform accountability on content moderation and make this information available to the public.
Beyond Content Removal: The Role of Prevention
In this sense, monitoring tools play an important role in facilitating content removal. By making citizens more aware of emerging trends, these tools may also encourage them to report hate content to platforms when it potentially violates platform rules. Content removal is therefore essential, as it reduces the visibility of harmful material and may limit immediate damage, particularly when content incites hatred or violence.
However, the development of official monitoring tools supports a highly complex task that is difficult to automate. Technology alone cannot effectively capture all forms of online hate: those creating hateful content can also be highly sophisticated in evading monitoring. Moreover, these technologies must be tested for bias to protect people from discrimination and prevent undue restrictions on freedom of expression.
But what happens when harmful content cannot be detected, does not constitute a criminal offence or an administrative violation, or does not infringe the EU’s Digital Services Act? Not all content can or should be addressed through takedowns alone, as different kinds of content require different responses.
Content removal, therefore, is not a complete solution on its own. Monitoring tools can offer significant potential for civil society by enabling a better understanding of the types of hate being produced and how they spread. This knowledge can support the development of counter-narratives and the promotion of critical thinking. Active civil society involvement is essential, with NGOs contributing to responses and helping to ensure that governance approaches are inclusive and sensitive to different contexts. Projects such as ECLIPSE focus on preventive approaches, using education and counter-narratives to build resilience against online hate and reduce the persuasive power of harmful content.
Authors: Lucía Alonso-Pérez & Paula Cano-Cruz
Links
Pérez Colomé, J: España se coloca en el centro de la lucha mundial contra el odio online: “Los grupos xenófobos están bien asesorados” (May, 2026) https://elpais.com/tecnologia/2026-05-01/espana-se-coloca-en-el-centro-de-la-lucha-mundial-contra-el-odio-online-los-grupos-xenofobos-estan-bien-asesorados.html
Keywords
Online hate; hate speech monitoring; content moderation; algorithmic amplification; counter-narratives; critical thinking; digital resilience.
[1] OHCHR: Special considerations for hate speech, available at: https://www.unhcr.org/handbooks/informationintegrity/understanding-challenge/special-considerations-hate-speech
[2] OBERAXE: Annual Report on Hate Speech Monitoring on Social Networking Site 2024, P. 16, available at: https://www.inclusion.gob.es/documents/6602794/7031492/Monitorizacion-discurso-del-odio_2024_EN+%281%29.pdf/c97a9449-76a2-feb0-72e0-d61b57da57a6?t=1758273010249
[3]El País, “El odio en redes se disparó durante la crisis de Torre Pacheco y las plataformas no estuvieron a la altura” (5 August 2025), available at: https://elpais.com/expres/2025-08-05/el-odio-en-redes-se-disparo-durante-la-crisis-de-torre-pacheco-y-las-plataformas-no-estuvieron-a-la-altura.html
[4]In other countries, see for example Amnesty International: X’s design and policy choices created fertile ground for inflammatory, racist narratives targeting Muslims and migrants following Southport attack in the UK (2025) , available at: https://www.amnesty.org/en/latest/news/2025/08/xs-design-and-policies/
[5] Jordi Pérez Colomé, “España se coloca en el centro de la lucha mundial contra el odio online: ‘Los grupos xenófobos están bien asesorados’,” El País, 1 May 2026, available at: https://elpais.com/tecnologia/2026-05-01/espana-se-coloca-en-el-centro-de-la-lucha-mundial-contra-el-odio-online-los-grupos-xenofobos-estan-bien-asesorados.html?utm_source=chatgpt.com
[6]FARO regular monitoring reports can be consulted on the website of OBERAXE at https://www.inclusion.gob.es/ca/web/oberaxe/monitorizacion-discurso-de-odio-ilegal-en-linea
[7] LALIGA (officially Liga Nacional de Fútbol Profesional) is the private sports association in Spain responsible for organising the two top tiers of professional men’s football.
[8] Available at: https://www.inclusion.gob.es/ca/web/oberaxe/hodio