ITU project to monitor illegal hate speech on social media platforms
Social media platforms are required by law to remove illegal content – such as hate speech – but in Denmark, monitoring of the platforms ceased in 2018. A new ITU lead project aims to monitor platforms for instances of hate speech in Danish and report to the EU.
Leon DerczynskiComputer Science Departmentethicssocial media
Written 27 February, 2023 23:00 by Theis Duelund Jensen
Hate speech and other types of content that incite violence against others on social media platforms is illegal in the European Union. Social media platforms, the majority of which are based outside of the EU, have very little incentive to tackle the problem, which is why most EU countries have independent organizations monitoring social media content to ensure that the platforms follow the law.
However, in Denmark no official monitoring of social media content for hate speech has been conducted since 2018. The result is that hate speech has been allowed to thrive on Danish language versions of social media platforms. A new ITU branch of the Safenet-project funded by EU's "Citizens, Equality, Rights and Values Programme" (CERV) seeks to remedy that. The project’s main goal is to monitor, collect data, analyse, and understand the moderation of hate speech done by the IT platforms; and to propose an adapted advocacy for EC institutions in order to tackle online phenomena.
“Most countries in the EU, as well as Switzerland and Norway, have people who will check whether the platforms follow the law. Denmark did until 2018. Then this activity was dropped. Since then, the law has still applied, but nobody is monitoring the platforms which means there is no enforcement, and there is no motivation for the platforms to do better,” says Associate Professor at the IT University of Copenhagen, Leon Derczynski, who leads the project.
“Once the content is reported, the platform is obligated to remove it, however, if the platform ignores this and leaves the content there, in Denmark, nothing will happen unless someone reports it directly to the police. There is no motivation for the platforms to comply with the law.”
Leon Derczynski’s area of expertise is Natural Language Processing and Machine Learning. He conducts research in online harm and hate speech and has previously been involved in developing an algorithm that detects hate speech directed at women on social media platforms.
Specifically, the project at ITU will, informed by the procedures of similar initiatives in other EU countries, conduct different monitoring exercises to ensure that social media platforms are in compliance with EU law:
“The exercises can be done in one of two ways. One way is to announce to a platform ahead of time that we are monitoring them. Another way is to do stealth monitoring which is kind of like a surprise inspection. The feedback goes through a special channel to the platforms, and if they do not remove the content, we will report it to the EU,” says Leon Derczynski.
The two-year project officially started on January 1, and according to the researcher it will be particularly interesting to see how hate speech manifests on TikTok which has not previously been subject to monitoring in a Danish language context.
“EU law is quite clear on hate speech but without monitoring, the platforms have had no incentive to address the problem,” says Leon Derczynski. “This project means we will get a better understanding what is really going on and start getting the platforms to clear up their act in Denmark."
Theis Duelund Jensen, Press Officer, tel: 2555 0447, email: thej@itu.dk