Monthly Report June 2025
Data for June was based on 3,222,802 messages in 11 languages, across 6 social media platforms: Reddit, X, 4chan, Gab, YouTube, and Facebook.
Content warning: Presented data may contain disturbing language related to online hate speech.
Average toxicity
In June, the overall average toxicity level was approximately 0.215. Toxicity remained largely stable throughout the month, fluctuating between 0.20 and 0.22. The highest values (0.22) were observed consistently in the first half and again at the end of the month, while a slight dip to 0.20 occurred between June 24 and June 26 before returning to previous levels.
Baseline channel analysis
Throughout June, toxicity levels across the monitored baselines remained relatively similar to the previous months, with some notable fluctuations. The anti-LGBTQ+ baseline showed a distinct peak on June 23rd, reaching a toxicity score of 0.35—the highest value recorded for this narrative during the month. This spike stands out from the otherwise consistent trend for LGBTQ+ content, which mostly hovered between 0.28 and 0.31. In contrast, the sexism and anti-refugee baselines maintained the lowest average toxicity levels, remaining stable between 0.18 and 0.24 throughout the month.
The antisemitism and anti-Muslim baselines exhibited only slight daily variation, consistently ranging between 0.31 and 0.34, and 0.26 to 0.29 respectively. Meanwhile, the anti-Roma narrative displayed more volatility, peaking at 0.33 on June 19th, with a toxicity average of 0.29.
Baseline Sexism
Volume: 356,541 posts identified as toxic content.
Platforms: Most posts originated from X (77%), with TikTok (12%) and Reddit (9%) also contributing significantly.
Toxicity: The average toxicity score was 0.19. Posts with extreme toxicity (0.8 or higher) totalled 19,639, predominantly on Gab.
Themes:
Sexism: 50% - the dominant hate category
Violence: 13% - concerning levels of violent rhetoric
Common toxic keywords included: slut, whore, bitch, sluts, and freak (English language terms with derogatory meanings targeting women).
Key Insight: This represents a predominantly sexist hate speech environment where misogynistic language is systematically weaponised, with the most extreme toxicity concentrated on alternative platforms like Gab, suggesting potential migration of the most harmful content away from mainstream social media moderation.
Baseline Anti-LGBTQ+
Volume: 103,702 posts identified as toxic content.
Platforms: Most posts originated from X (83%), with Reddit (9%) and TikTok (6%) also contributing significantly.
Toxicity: The average toxicity score was 0.30. Posts with extreme toxicity (0.8 or higher) totaled 6,034, predominantly on Gab.
Themes:
Sexism: 77% - the dominant hate category
Violence: 10% - notable levels of violent rhetoric
Common toxic keywords included: slut, whore, sluts, bitch (English derogatory terms), and شاذ (Arabic: "shādh" meaning "deviant" or "pervert").
Key Insight: This represents a predominantly sexist hate speech environment with strong anti-LGBTQ+ undertones, where misogynistic language intersects with homophobic and transphobic slurs, indicating intersectional targeting of gender and sexual minorities, with the most extreme content concentrated on alternative platforms like Gab.
Baseline Anti-Muslim
Volume: 166,330 posts identified as toxic content.
Platforms: Most posts originated from X (89%), with TikTok (6%) and Reddit (3%) also contributing significantly.
Toxicity: The average toxicity score was 0.28. Posts with extreme toxicity (0.8 or higher) totaled 14,601, predominantly on Reddit.
Themes:
Religion: 77% - the dominant hate category
Violence: 21% - significant levels of violent rhetoric
Common toxic keywords included: slut, whore, sluts (English derogatory terms), and moros, moro (Spanish: derogatory terms historically used against Muslims, meaning "Moor").
Key Insight: This represents a predominantly anti-Muslim hate speech environment where religious targeting intersects with historical ethnic slurs and misogynistic language, indicating intersectional hatred that combines Islamophobia with gendered attacks, with the most extreme toxicity concentrated on Reddit's discussion-based format.
Baseline Anti-Refugee/Migrants
Volume: 38,451 posts identified as toxic content.
Platforms: Most posts originated from X (87%), with Reddit (5%) and TikTok (4%) also contributing significantly.
Toxicity: The average toxicity score was 0.22. Posts with extreme toxicity (0.8 or higher) totaled 1,196, predominantly on Reddit.
Themes:
Politics: 62% - the dominant hate category
Violence: 16% - notable levels of violent rhetoric
Common toxic keywords included: slut, whore, bitch (English derogatory terms), حد (Arabic: "ḥadd" meaning "limit/boundary," potentially referring to Islamic law), and fachkräfte (German: "skilled workers," used sarcastically regarding refugees/migrants).
Key Insight: This represents a politically-driven hate speech environment targeting refugees and migrants, where anti-immigration sentiment intersects with misogynistic language and sarcastic references to skilled worker policies, indicating intersectional hatred that combines xenophobia with gendered attacks, with the most extreme content concentrated on Reddit's discussion platforms.
Baseline Antisemitism
Volume: 187,241 posts identified as toxic content.
Platforms: Most posts originated from X (92%), with Reddit (3%) and TikTok (3%) also contributing significantly.
Toxicity: The average toxicity score was 0.33. Posts with extreme toxicity (0.8 or higher) totaled 11,159, predominantly on Gab.
Themes:
Racism: 75% - the dominant hate category
Violence: 22% - significant levels of violent rhetoric
Common toxic keywords included: slut, whore, sluts, bitch (English derogatory terms), and الصهاينة (Arabic: "al-sahāyna" meaning "the Zionists").
Key Insight: This represents a heavily racialised hate speech environment with concerning levels of violent rhetoric, where antisemitic targeting intersects with misogynistic language patterns, and the most extreme toxicity migrates to alternative platforms like Gab beyond mainstream social media oversight.
Baseline Roma
Volume: 18,085 posts identified as toxic content.
Platforms: Most posts originated from X (84%), with Reddit (9%) and TikTok (4%) also contributing significantly.
Toxicity: The average toxicity score was 0.29. Posts with extreme toxicity (0.8 or higher) totalled 1,600, predominantly on Gab.
Themes:
Racism: 84% - the dominant hate category
Violence: 10% - notable levels of violent rhetoric
Common toxic keywords included: slut, whore, sluts, bitch (English derogatory terms), and gitanos (Spanish: derogatory term for Roma people, meaning "gypsies").
Key Insight: This represents a predominantly anti-Roma hate speech environment where racial targeting intersects with misogynistic language, indicating intersectional hatred that combines anti-Roma racism with gendered attacks, with the most extreme content migrating to alternative platforms like Gab beyond mainstream social media moderation.
This graph illustrates the distribution of toxicity levels—Neutral, Low, Medium, and High—across our monitored baselines. The anti-Muslim and antisemitism datasets had the highest proportion of highly toxic messages, at 2.3% and 2.0% respectively. Notably, the antisemitism baseline also had the lowest share of neutral (27.7%) toxicity posts, suggesting particularly polarised or hostile discussions. In contrast, the sexism and anti-refugee baselines showed the least toxic behaviour, with the lowest rates of highly toxic messages—0.7% and 0.6% respectively—and the highest share of neutral content, at 57.6% and 46.4%.
Hate speech by category
The hate speech categorisation data highlights clear thematic overlaps across the six baselines. Antisemitic content is most strongly associated with racism (75.94%), followed by religion (47.37%) and politics (41.96%). Anti-Muslim discourse is similarly rooted in religion (69.70%), with significant intersections with racism (43.11%) and politics (31.53%). Anti-LGBTQ+ content is largely framed through sexism (60.72%) and obscenity (11.84%), while sexist narratives also intersect with politics (17.49%) and ridicule (11.82%).
Anti-refugee content is primarily political (46.21%) and racist (37.58%), with lower levels of overt toxicity compared to other baselines. Meanwhile, anti-Roma narratives show strong associations with racism (51.14%) and contempt (8.83%), reflecting entrenched bias. Overall, racism, politics, religion, and sexism remain the dominant lenses through which hate speech manifests, with antisemitic and anti-Muslim posts containing the highest proportions of threats, pointing to more extreme forms of online hostility toward these groups.