This monthly report examines the key digital narratives that have emerged in our baseline channels as significant vectors for hate speech, tracking their evolution, reach and potential impact on social discourse.

By mapping these baseline channels, we aim to offer insights that can inform policy-making, platform moderation strategies, and broader efforts to combat online hatred. Our analysis encompasses a diverse range of digital platforms, including social media platforms, image boards, forums, and comment sections. Through sophisticated analytical methodologies, we aim to provide a nuanced understanding of the current state of online hate speech in the European digital ecosystem.

Baseline channel analysis

Baseline Sexism

The data reveals that during this reporting period, a total of 363,498 posts were identified as containing hate speech.

*Sexism toxicity timeline*
Toxicity reflects the average score of analysed posts, rated from 0 (safe) to 1 (extremely hateful)

Let's take a closer look at the sources of this hateful content. The majority, around 55%, was collected from the social media platform X. YouTube and TikTok also contributed notable shares at 20% and 18% respectively. This distribution highlights that hate speech is pervasive across multiple digital environments.

*Percent of total posts by toxicity*
*Shows the percent of messages in subscribed channels that fall into each toxicity level (safe, low, medium, and high).*

Another important metric tracked in the dashboard is the "toxicity score". The overall average toxicity score for the 363,498 posts was 0.17, on a scale where 1.0 represents the highest level of toxicity.

While the average score is relatively low, the dashboard identified 2,109 posts that have a toxicity score of 0.8 or higher. Further analysis reveals that these severely toxic posts are more frequently found on the online imageboard 4chan.

Exploring the thematic aspects of hate speech, the data reveals that approximately 12% of the messages focus on religion, 9% center on politics, and 27% on violence. This suggests that divisive topics around faith and ideology can often serve as breeding grounds for the propagation of hateful views.

*Visualisation of the intersections between different categories in the Sexism baseline*

The most common toxic keywords include racial slurs such as "niggers" and "nigger", as well as derogatory terms like "moron", "faggot", and "scum". The prevalence of these dehumanising and discriminatory terms highlights the pervasive nature of racism, homophobia, and other forms of bigotry underlying much of the online hate.

Baseline Anti-LGBTQ+

Our dataset reveals that this channel contains 36,053 posts identified as exhibiting anti-LGBTQ+ sentiments and rhetoric. While the overall volume may be lower compared to some other categories, the impact of such hateful content can be particularly damaging, contributing to the marginalisation and oppression of LGBTQ+ individuals both online and in the real world.

*Anti-LGBTQ+ toxicity timeline*

Toxicity reflects the average score of analysed posts, rated from 0 (safe) to 1 (extremely hateful)

Examining the sources of this hateful content, we see that the majority, around 60%, was collected from the social media platform X. YouTube and TikTok also contributed notable shares at 18% and 12% respectively.

The "toxicity score" metric is particularly concerning in this context, with the overall average for the 36,053 posts standing at 0.32, on a scale where 1.0 represents the highest level of toxicity. Even more concerning is the identification of 552 posts that have a toxicity score of 0.8 or higher. These highly toxic messages are particularly alarming especially egregious, containing the most extreme and dehumanising forms of anti-LGBTQ+ rhetoric.

Further analysis reveals that these severely toxic posts are more frequently found on the alternative social media platform Gab, once again highlighting the need for increased moderation and content regulation on emerging digital channels that may serve as breeding grounds for such hateful content.

Delving deeper into the thematic aspects of the anti-LGBTQ+ hate speech, the data indicates that around 18% of the messages also involve underlying sexist attitudes and biases. This intersection of homophobia and misogyny underscores the complex and interconnected nature of the discrimination faced by LGBTQ+ individuals.

Additionally, a concerning 25% of the posts contain references to violence., which is extremely troubling and requires immediate intervention. Some of the common toxic keywords identified in these messages include derogatory terms like "faggot", "lesbo", "niggers", "gay and retarded", and "fucking gay".

The prevalence of such dehumanising and discriminatory language is a stark reminder of the urgent need to address the underlying homophobia, transphobia, and other forms of bigotry that fuel this online hate. By shedding light on these trends, this baseline analysis can help guide policy decisions, platform moderation strategies, and grassroots efforts to create more inclusive, respectful digital communities where LGBTQ+ individuals can freely and safely express their identities.

*Visualisation of the intersections between different categories in the Anti-LGBTQ+ baseline*

Baseline Anti-Muslim

We also uncovered insights into the proliferation of anti-Muslim hate speech within the digital landscape. The data reveals that this channel contains 72,991 posts identified as exhibiting anti-Muslim sentiments and rhetoric.

Examining the sources of this hateful content, we see that the majority, around 52%, was collected from the social media platform X. YouTube and TikTok also contributed notable shares at 33% and 12% respectively.The "toxicity score" metric reveals an overall average of 0.23 across the 72,991 posts analysed.

While this average may seem relatively low, the dashboard also identifies 502 posts that have a toxicity score of 0.8 or higher.

Further analysis reveals that these severely toxic posts are more frequently found on the alternative online forum 4chan.

Delving deeper into the thematic aspects of the anti-Muslim hate speech, the data indicates that around 30% of the messages involve religious themes, and 9% also exhibit underlying racist undertones. This intersection of religious intolerance and racism highlights the compounded discrimination faced by Muslim individuals.

Additionally, a concerning 48% of the posts contain references to violence. Some of the common toxic keywords identified in these messages include derogatory terms like "kuffar", "evil spirits", "niggers", "inbred", and "islam is a cancer".

Baseline Anti-Refugee/Migrants

This analysis sheds light on the pervasive and concerning nature of xenophobic content circulating online. The data reveals that this channel contains 14,948 posts identified as exhibiting anti-refugee and anti-migrant sentiments and rhetoric.

Examining the sources of this hateful content, we see that the majority, around 65%, was collected from the social media platform X. YouTube and TikTok also contributed notable shares at 24% and 4% respectively..

The toxicity score shows an overall average for the 14,948 posts standing at 0.24, on a scale where 1.0 represents the highest level of toxicity. 111 posts have a toxicity score of 0.8 or higher.

Further analysis reveals that these severely toxic posts are more frequently found on the alternative online forum 4chan.

Delving deeper into the thematic aspects of the anti-refugee and anti-migrant hate speech, the data indicates that around 16% of the messages also involve political themes, and 10% exhibit underlying racist undertones. This intersection of xenophobia, politics, and racism underscores the complex and interconnected nature of the discrimination faced by refugee and migrant populations.

Additionally, a concerning 70% of the posts contain references to violence. Some of the common toxic keywords identified in these messages include derogatory terms like "migrants", "#endtimes", "tnd", "niggers", and "scum".

Baseline Antisemitism

We also uncovered insights into the proliferation of antisemitic hate speech within the digital landscape. The data reveals that this channel contains 100,836 posts identified as exhibiting antisemitic sentiments and rhetoric. Examining the sources of this hateful content, we see that the majority, around 62%, was collected from the social media platform X. YouTube and the alternative online forum 4chan also contributed notable shares at 24% and 8% respectively.

*Antisemitism toxicity timeline*
Toxicity reflects the average score of analysed posts, rated from 0 (safe) to 1 (extremely hateful)

The toxicity score shows an overall average for the 100,836 posts standing at 0.29. We identified 2,229 posts that have a toxicity score of 0.8 or higher. Further analysis reveals that these severely toxic posts are more frequently found on the 4chan platform.

Delving deeper into the thematic aspects of the antisemitic hate speech, the data indicates that around 16% of the messages also exhibit underlying racist undertones, and 12% involve political themes. This intersection of antisemitism, racism, and politics highlights the complex and interconnected nature of the discrimination faced by Jewish individuals and communities.

Additionally, a concerning 52% of the posts contain references to violence.Some of the common toxic keywords identified in these messages include racial slurs like "niggers", as well as explicit, violent language such as "kill jews", "kike", "kikes", and "faggot".

*Visualisation of the intersections between different categories in the Antisemitism baseline*

Baseline Roma

The Roma people, also known as Romani, have long faced systemic discrimination and marginalization, and this online hate speech serves to further perpetuate these harmful biases. The dashboard indicates that this "Baseline Anti-Roma" channel contains 2,246 posts that exhibit anti-Roma sentiment and rhetoric. While this volume may seem relatively lower compared to some other categories, it is crucial to recognise that even a smaller number of hateful posts can have a significant and insidious impact, normalising and reinforcing prejudiced attitudes toward the Roma community.

*Anti Roma toxicity timeline*
Toxicity reflects the average score of analysed posts, rated from 0 (safe) to 1 (extremely hateful)

The data shows that the majority, around 41%, was collected from the video-sharing platform YouTube. X and TikTok also contributed notable shares at 26% and 17% respectively. This distribution suggests that anti-Roma hate speech is present across multiple social media ecosystems, requiring coordinated efforts to address the problem.

One particularly troubling metric is the toxicity score. For the 2,246 anti-Roma posts, the overall average toxicity score was 0.33, on a scale where 1.0 represents the highest level of toxicity. While this average may seem relatively low, the dashboard also identifies 56 posts that have a toxicity score of 0.8 or higher.

Further analysis reveals that these severely toxic posts are more frequently found on the alternative social media platform Gab.

Delving deeper into the thematic aspects of the anti-Roma hate speech, the data indicates that around 16% of the messages also exhibit underlying racist undertones. This intersection of anti-Roma bias and broader racism highlights the compounded, intersectional nature of the discrimination faced by the Roma people both online and in the physical world.

Additionally, 35% of the posts contain references to violence. Some of the common toxic keywords identified in these messages include the disturbing use of the Nazi swastika symbol ("卐"), as well as other derogatory terms like "luciferian", "liar", "niggers", and "nigger".

Monthly Report November 2024