Nowadays, recommender systems are increasingly important to various users’ digital interactions across top platforms. Furthermore, this includes online marketplaces, entertainment services, social media outlets, and e-commerce platforms. While the appeal of these systems is broad, they are susceptible to substantial biases generating unfair outcomes and lower content diversity. In this survey paper, we present reasons, sort them into sources of bias and types of bias in recommender systems, and discuss how different kinds of biases appear in recommender systems in real systems. It also identifies biases caused by recommendations, resulting in unfairness, transparency, accountability, and privacy issues. The paper also walks through current mitigation strategies. This comprises debiasing algorithms, fairness-aware models, and user-centric approaches to alleviate fictitious biases. Drawing on widely used platforms like Netflix, YouTube, and Spotify for case studies, the survey depicts the practical impact of biased recommendations and evaluates the shortcomings of existing solutions. Additionally, the paper reveals open research challenges and indicates possible ways of improving algorithmic fairness, transparency, and users’ privacy in recommender systems. This work contributes to a more extensive ongoing discussion about the ethical use of AI technologies, particularly about responsible innovation during the development of recommender systems.