A recent investigation has exposed concerning failures in content moderation by major social media platforms Meta and X ahead of Germany's federal election on February 23. According to research by corporate responsibility nonprofit Eko, both platforms approved advertisements containing violent anti-Muslim and antisemitic content.
The study found that X (formerly Twitter) approved all 10 test ads containing hate speech, while Meta approved 5 out of 10 similar submissions for Facebook and Instagram. The approved content included violent messaging targeting religious and ethnic minorities, with some ads calling for extreme actions against Muslim refugees and Jewish communities.
Among the approved advertisements were those comparing refugees to "vermin," promoting violence against religious buildings, and spreading antisemitic conspiracy theories. The ads were accompanied by AI-generated images that went unlabeled, despite platform policies requiring disclosure of artificial imagery in political advertising.
The findings raise questions about both platforms' commitment to their stated content moderation policies and compliance with the European Union's Digital Services Act (DSA). The DSA requires platforms to protect against illegal content and systemic risks, with potential penalties of up to 6% of global annual turnover for violations.
Eko's researchers prevented the test ads from actually running to protect users. They have shared their findings with the European Commission, which currently has ongoing DSA investigations into both Meta and X regarding election security and illegal content concerns.
The timing is particularly sensitive given Germany's upcoming federal election, where immigration has become a central campaign issue. X's owner Elon Musk has already faced criticism for publicly supporting the far-right AfD party in December 2023.
The research suggests neither platform is adequately enforcing their own hate speech policies for advertisements. Eko representatives noted that Meta's AI-driven moderation systems appear "fundamentally broken" despite new EU regulations, while X's approval of all test ads raises questions about whether any meaningful ad review process exists on the platform.
These findings come as part of broader concerns about social media's impact on democratic processes, with other recent studies suggesting algorithmic biases favoring certain political content on major platforms.