What’s happening online as the Commission cracks down on hate speech?

The European Commission against Racism and Intolerance has called on Sweden, Portugal, Croatia and Latvia to take stronger action against hate speech, predominantly targeting migrants, Roma, LGBTQ+ and Black people.

This call comes as an average medium level of online toxicity has been recorded since the beginning of 2025, according to the European Observatory of Online Hate (EOOH).

Online toxicity encompasses rude, aggressive, and degrading attitudes and behaviour exhibited on online platforms. It can range from an excessive use of profanity to outright hate speech.

The EOOH scores toxicity in four different categories from safe (score of 0) to high (score between 0.8 and 1).

The toxicity score is determined by lists of hateful and other problematic words and phrases. These lists were compiled from social media, and each entry was assigned a toxicity level along with specific categories of toxicity, such as sexism, racism, and others.

In April 2025, the observatory recorded the highest level of toxicity seen so far this year, with a score of 0.22 out of 1.

This score falls into the medium category, in which messages are found to be toxic without being extreme.

In May this year, antisemitism maintained the highest average toxicity score at 0.34, with 88% of the posts originating from X (formerly Twitter).

The European Observatory of Online Hate analysed more than 2.5 million messages in 11 languages, across six social media platforms in May.

After antisemitic comments, anti-Roma content was the second-most common kind of online hate speech. A score of 0.30 highlights this community continues to be targeted by online users.

Most posts originated from X (82%), with Reddit (7%) and TikTok (4%) also contributing significantly.

Anti-LGBTQ+ and anti-Muslim narratives both registered significant toxicity levels at 0.29 and 0.28, respectively, while anti-refugee content scored 0.23.

This toxic content was mainly shared on X, at 81% and 87%, respectively.

Although sexist content recorded the lowest average toxicity score at 0.19, these insults still represented the highest volume of toxic posts, with the observatory noting almost 3 million examples.

Misogynistic language dominated the most harmful discourse.

This figure is over three times the number of anti-Roma posts and more than double that of posts targeting LGBTQ+ individuals.