Artificial intelligence (AI) systems constantly pose unknown challenges to fundamental rights institutions. However, this research focuses specifically on freedom of religion and the requirement of non-discrimination. The nature of machine learning-based algorithms is lack of transparency and the amplification of human and social biases in their training data. The latter is called algorithmic bias, which can result in systemic and hidden discrimination. Large language and image-generating AI models unjustifiably associate religious minorities with negative stereotypes, violence or terrorism. Meanwhile, content moderation algorithms may disproportionately classify legitimate religious expressions as offensive, thereby undermining the external manifestation of religious practice. Given that the violations outlined above are caused by private parties and platforms, the classic vertical protection of fundamental rights between the state and the individual is insufficient on its own. Consequently, this research examines the possibilities of extending constitutional legal protection to private law relationships, specifically to legal relationships between platform users and employers and employees, through the doctrinal framework of the horizontal effect (Drittwirkung) of fundamental rights. The research focuses on the case law of the ECtHR and the regulatory instruments of the European Union, the Digital Services Act and the AI Act.