Monthly Report December 2025
Data for December 2025 was based on 13.7M messages in 25 languages across 12 social media platforms such as X, YouTube, TikTok, Threads and Instagram.
Content warning: Presented data may contain disturbing language related to online hate speech.
Average toxicity
Throughout December, average toxicity levels remained largely stable around 0.20, with a brief uptick to 0.22 in mid-month before settling back to approximately 0.21 toward the end of the month.
Baseline channel analysis
Throughout December 2025, toxicity levels across all baselines remained relatively stable, with clear differences in intensity between categories. Antisemitism consistently recorded the highest values, fluctuating between 0.28 and 0.31, with a slight increase half December and at the end of the month. Anti-LGBTQ+ content followed, showing moderate and steady levels that gradually edged upward toward the end of the month, peaking at 0.29. Anti-Muslim toxicity remained lower and stable overall, ranging between 0.22 and 0.25, with minor increases around mid-December. Anti-refugee narratives displayed limited variation, generally staying between 0.21 and 0.23 throughout the period. Anti-Roma toxicity rose from around 0.15 early in the month to an average of 0.19 in the second half. Sexism remained the least toxic baseline overall, holding consistently at low levels around 0.16-0.17 with minimal fluctuation.
Baseline Sexism
Volume: 5.1M of 53.4M posts (9.5%) in this channel were identified as toxic.
Platforms: Most data was observed on X (52%), and YouTube (35%).
Languages: Posts are mostly written in English (51%), and Arabic (11%).
Toxicity Analysis: The average toxicity score was 0.16. Posts with very high toxicity (≥0.8) totaled 217k, and are most prominent on Chirpwire (12%).
Toxicity Categories:
49% of toxic posts were categorised as Sexism.
35% of those toxic posts contained violent rhetoric.
Commonly used words in those toxic posts included: women, woman, الله, stupid, girl, femmes, femme, الصهاينة, bitch, homme
Key Findings: The provided social media posts reveal a deeply polarized and often aggressive discourse surrounding various social and political issues, with a recurring theme of animosity and distrust directed towards specific groups. Several posts express strong anti-immigrant sentiments, using derogatory language and associating immigrants with crime and societal decay. There are also numerous instances of misogynistic language and threats directed at women, often in response to discussions about gender equality or feminist movements. Political figures and ideologies are frequently targeted with insults and accusations, ranging from corruption to fascism. Religious and ethnic groups are also subjected to hateful rhetoric and stereotypes. The language used is consistently inflammatory, employing slurs, profanity, and violent imagery. Specific locations and events are rarely mentioned, with the focus being on broad generalizations and attacks against perceived adversaries. The overall tone is one of anger, frustration, and a lack of constructive dialogue, reflecting a society grappling with significant divisions and a breakdown in civil discourse.
Selected News items
• Radiografía de los sintecho en Cataluña: se multiplica por tres el número de mujeres sin hogar
• Morena y Verde se dividen en SLP por una reforma que beneficia a la esposa del gobernador
• Esas chistorras, fontaneras, aerolíneas y braguetas de las que usted me habla
• Marco Borsato vrijgesproken van ontucht met vijftienjarig meisje
Baseline Anti-LGBTQ+
Volume: 1M of 7M posts (14.9%) in this channel were identified as toxic.
Platforms: Most data was observed on X (72%), and YouTube (17%).
Languages: Posts are mostly written in English (54%), and Spanish (10%).
Toxicity Analysis: The average toxicity score was 0.27. Posts with very high toxicity (≥0.8) totaled 38k, and are most prominent on Gab (3%).
Toxicity Categories:
74% of toxic posts were categorised as Sexism.
17% of those toxic posts contained violent rhetoric.
Commonly used words in those toxic posts included: homo, women, schwule, lésbica, lesbiana, lesbica, retarded, lesben, queer, travesti
Key Findings: The provided posts contain a significant amount of hateful and derogatory language directed towards individuals based on their perceived sexual orientation and gender identity. Several posts from Spain and Colombia use extremely offensive terms to insult political figures and opponents, often linking them to homosexuality and gender non-conformity in a derogatory manner. For instance, one post from Colombia repeatedly calls a political figure a "travesti HP" and "anormal de mierda." Similarly, posts from Germany and France use slurs like "Schwuchtel" and "tarlouze" while expressing disdain for certain political groups or individuals. There are also instances of religious condemnation, with some posts linking homosexuality to sin and damnation. Some users express concern about children becoming homosexual or transgender, while others make disturbing claims about "children from the penis of a Muslim with a gay parodist." The posts also reveal a deep distrust and animosity towards certain political parties and ideologies, with accusations of them promoting or being associated with individuals who engage in harmful behaviors. The language used is consistently aggressive, dehumanizing, and often includes threats and violent imagery. The prevalence of slurs and hateful rhetoric across multiple languages and regions highlights a widespread issue of intolerance and prejudice.
Selected News items
• Irão e Egipto foram seleccionados para Jogo do Orgulho LGBT do Mundial. Não estão felizes
• Seattle setzt auf Pride Match trotz Iran gegen Ägypten
• WM 2026: Iran will Pride-Spiel bei WM unbedingt verhindern
• Fußball-Weltmeisterschaft 2026: Iran will Pride-Spiel in den USA gegen Ägypten unbedingt verhindern
• Iran will Pride-Spiel bei WM "unbedingt verhindern" | Heute.at
Baseline Anti-Muslim
Volume: 2.3M of 13.8M posts (16.6%) in this channel were identified as toxic.
Platforms: Most data was observed on X (72%), and YouTube (20%).
Languages: Posts are mostly written in English (51%), and Arabic (16%).
Toxicity Analysis: The average toxicity score was 0.24. Posts with very high toxicity (≥0.8) totaled 163k, and are most prominent on Chirpwire (13%).
Toxicity Categories:
81% of toxic posts were categorised as Religion.
37% of those toxic posts contained violent rhetoric.
Commonly used words in those toxic posts included: muslims, muslim, moros, islam, islamic, moro, الله, الإسلام, المسلمين, quran
Key Findings: The social media posts reveal a strong undercurrent of anti-Muslim sentiment, often expressed through derogatory language and generalizations. Several posts target individuals or groups perceived as "traitors" for their perceived support of or integration with Muslim communities. For instance, one post in Spanish criticizes someone for "handing over your homeland to the Moors and Berbers." Another, in Dutch, refers to Muslims as "dirty Muslim scum" and attributes low IQ to them. French posts also contain aggressive language, with one user telling someone to "shut your mouth, you collaborator" and another to "go blow yourself up in Gaza." German posts include a vivid, albeit offensive, anecdote about a street preacher and a Muslim woman. Arabic posts also reflect animosity, with one stating, "Don't lie, Daesh is not practicing Islam," and another expressing a desire for Arabs to be killed. Spanish posts continue this trend, with one user stating, "There never existed Palestine and the fucking Muslims started the mess in 1948." The sentiment extends to accusations of hypocrisy and calls for expulsion, as seen in a Dutch post wishing for "dirty Muslim scum" to be "kicked out of the country." Some posts also connect Muslims with negative stereotypes, such as low intelligence or a lack of values. The overall tone is one of anger, distrust, and a desire to exclude or condemn Muslim individuals and the religion itself.
Selected News items
• Siria: Doi oameni au murit, după proteste alawite declanșate de un atac asupra unei moschei
• Nigeria : les États-Unis annoncent avoir mené des frappes meurtrières contre l'État islamique
• Donald Trump o "śmiertelnym ciosie" w Nigerii. Hegseth: to nie koniec
• Donald Trump grzmi po ataku ISIS! "Poważny odwet" za śmierć żołnierzy w Syrii
Baseline Anti-Refugee/Migrants
Volume: 436.8k of 3.8M posts (11.6%) in this channel were identified as toxic.
Platforms: Most data was observed on X (73%), and YouTube (16%).
Languages: Posts are mostly written in English (36%), Spanish (16%), and German (13%).
Toxicity Analysis: The average toxicity score was 0.22. Posts with very high toxicity (≥0.8) totaled 17k, and are most prominent on Gab (1%).
Toxicity Categories:
66% of toxic posts were categorised as Politics.
35% of those toxic posts contained violent rhetoric.
Commonly used words in those toxic posts included: refugees, migrant, لاجئ, migrante, asylanten, immigrati, deutschland, vluchtelingen, refugee, flüchtling
Key Findings: The social media posts reveal a strong undercurrent of anti-immigrant and anti-refugee sentiment across various European countries. Several posts specifically mention Germany, with one user expressing frustration about the country being "flooded with migrants, Ukrainians, Afghans and Syrians." Another post from Germany criticizes established parties for having "no problem with rapists, child molesters, drug dealers and murderers, if they are migrants." There are also mentions of specific locations within Europe, such as "England" and "Sweden," with users expressing concerns about crime and societal changes attributed to immigration. The posts frequently use derogatory terms and generalizations, labeling immigrants and refugees as "invaders," "criminals," "rapists," and "scroungers." Some posts also touch upon political figures and parties, accusing them of enabling or promoting policies that are seen as detrimental due to immigration. For instance, one post criticizes "established parties" in Germany, while another mentions "Starmer" in the context of immigration and crime in England. The language used is often aggressive and accusatory, reflecting a deep-seated anger and fear regarding the presence of non-European migrants and refugees.
Selected News items
• BKA-Lagebild: BKA meldet für 2024 sinkenden Anteil tatverdächtiger Zuwanderer
• Letland voltooit bouw ruim 280 kilometer lang hek op Lets-Russische grens
• Futura lei da deportação de imigrantes ilegais prevê prazo de detenção até ano e meio
• Los migrantes del instituto B9 de Badalona piden al Tribunal de Estrasburgo que frene su desalojo
Baseline Antisemitism
Volume: 2.2M of 11.2M posts (20.1%) in this channel were identified as toxic.
Platforms: Most data was observed on X (77%), and YouTube (14%).
Languages: Posts are mostly written in English (59%).
Toxicity Analysis: The average toxicity score was 0.30. Posts with very high toxicity (≥0.8) totaled 136k, and are most prominent on Gab (4%).
Toxicity Categories:
84% of toxic posts were categorised as Racism.
37% of those toxic posts contained violent rhetoric.
Commonly used words in those toxic posts included: jews, اليهود, jewish, israel, judíos, juifs, يهودي, israël, żydów, juden
Key Findings: The provided posts contain a significant amount of hateful and violent rhetoric directed towards Jewish people. Several posts explicitly call for the killing or extermination of Jews, using derogatory terms and dehumanising language. Some posts reference historical events, such as the Holocaust, in a distorted or celebratory manner. There are also accusations of Jews controlling global systems, such as media and finance, and engaging in harmful practices. The posts frequently link Jewish people to negative stereotypes and conspiracy theories. Some individuals express a desire for the destruction of Israel and the elimination of all Jewish people. The language used is often aggressive and threatening, with some posts directly inciting violence. The content spans multiple languages, indicating a widespread issue across different online communities. Specific locations like Gaza, Australia (Sydney), and Germany are mentioned in connection with these hateful sentiments. Some posts also target specific individuals or groups, accusing them of being Jewish or supporting Jewish interests, and subsequently directing abuse towards them.
Selected News items
• Vidio je zvjerstvo Hamasa na festivalu, a sad je ranjen u Australiji. "Puno je krvi"
• Tko su trenutno najpoznatiji otac i sin, monstrumi za koje zna cijeli svijet
• El presidente de Israel vincula a una "ola de antisemitismo" el tiroteo masivo en playa de Sídney
Baseline Roma
Volume: 290.1k of 1.8M posts (15.8%) in this channel were identified as toxic.
Platforms: Most data was observed on X (49%), and YouTube (40%).
Languages: Posts are mostly written in Russian (34%), English (17%), and Portuguese (12%).
Toxicity Analysis: The average toxicity score was 0.20. Posts with very high toxicity (≥0.8) totaled 17k, and are most prominent on Gab (6%).
Toxicity Categories:
85% of toxic posts were categorised as Racism.
15% of those toxic posts contained violent rhetoric.
Commonly used words in those toxic posts included: gitanos, gitano, gypsies, gypsy, zigeuner, roms, moros, cygan, roma, jews
Key Findings: The social media posts reveal a strong undercurrent of negative sentiment and prejudice directed towards the Roma community. Several posts use derogatory terms like "Zigeuner" (gypsy) in a derogatory manner, associating them with negative stereotypes such as greed, dirtiness, and criminality. For instance, one post states, "Nur Zigeuner sind ekelhaft gierig!" (Only gypsies are disgustingly greedy!). Another post expresses frustration with their presence, stating, "Entre gitanos y panchos el gimnasio de mi pueblo está para llenarlo de hormigón con la puertas cerradas." (Between gypsies and 'panchos', the gym in my town should be filled with concrete with the doors closed). There are also mentions of specific locations and events, such as a Christmas greeting from "Candidatura Carlos de Paz Moreno (Los Gitanos)" in Sevilla, Spain, and a musical performance titled "Caravana de gitanos" (Caravan of gypsies) in Pulianas. Some posts also touch upon the historical context, referencing the persecution of Roma people during the Nazi regime. The discourse frequently involves comparisons and conflation with other ethnic or racial groups, often in a negative light. The use of terms like "Negerkuss" (Negro kiss) and "Zigeunerschnitzel" (gypsy schnitzel) is also debated, with some defending their use and others criticizing them as offensive.
Selected News items
• Fetița care a supraviețuit Iadului a venit în România: Părinții au fost uciși în fața ei
• Narcos, esclavagismo e mentiras
• Dziadek "męża" Viki Gabor to "cygański król"! Oto kim jest i jak mieszka
• Un gran incendio devora varias infraviviendas en un poblado chabolista de Madrid
Sexism shows a high share of neutral content (61.2%), indicating that most discussions remain non-hostile. Low (20.7%) and medium (17.7%) toxicity are present but limited, while high-toxicity content is rare (0.4%).
Anti-Roma records the highest proportion of neutral posts (62.1%), suggesting relatively lower overall hostility. Low (18.1%) and medium (19.2%) toxicity levels appear comparable, with a small proportion of high toxicity (0.6%).
In the Anti-Muslim baseline, nearly half of the content is neutral (44.3%), but low (30.4%) and medium (24.5%) toxicity together make up a substantial share, indicating frequent expressions of hostility. High-toxicity content remains limited (0.9%).
Anti-Refugee displays a similar pattern, with 46.5% neutral content alongside low (31.4%) and medium (21.7%) toxicity, reflecting recurring but generally moderate antagonism. High toxicity is marginal (0.5%).
Anti-LGBTQ+ stands out for having the highest proportion of low-toxicity posts (46.3%). Neutral content accounts for 29.6%, while medium toxicity remains the third highest value (23.7%), after anti-muslims and antisemitism categories.
Antisemitism shows the most even distribution across neutral (32.7%), low (34%), and medium (32.2%) toxicity levels, indicating a broad spectrum of hostile expression. Although still low overall, it has the highest shares of high-toxicity posts (1.06%).
Hate speech by category
The data highlight clear patterns of intersectionality across baselines, showing how different forms of hate intersect with our baselines.
Antisemitic content is strongly shaped by racism (83.3%) and religion (41.7%), while anti-Muslim toxicity presents a mirrored figure; it is primarily shaped by religion (81.36%) and secondly by racism (50,48%). Both baselines also intersect substantially with political discourse (36.3% and 27.07%, respectively) and threats (21.46% and 22.92%), indicating highly ideological and confrontational narratives.
Anti-LGBTQ+ hate is primarily structured around sexism (64.53%), alongside politically-framed content (14.51%), pointing to a gendered and degrading mode of expression.
Sexism-related content similarly combines sexist framing (36.82%) with politics (14.02%) and threats (12.21%), reflecting persistent patterns of verbal hostility. Anti-refugee narratives stand out for their strong political (46.21%) and racial (38.48%) dimensions, frequently accompanied by threatening language (17%). At the same time, anti-Roma discourse remains predominantly racialised (28.49%), with smaller but consistent overlaps with politics and threats. Overall, racism, politics, religion, and threats emerge as recurring cross-cutting elements, illustrating how toxic language often operates through multiple, reinforcing dimensions rather than in isolation.