This chapter lays the theoretical foundation by reviewing key concepts and tools from NLP relevant to fake news and hate speech detection. It overviews the NLP workflows, such as text classification and language modeling (Sect. 2.1). Evaluation metrics—like F1-score, ROUGE, and BERTScore—are discussed for both classification and generation tasks (Sect. 2.2). The chapter then explores Transfer Learning (TL), differentiating inductive, and transductive approaches, with an emphasis on transformer-based models (e.g., BERT, RoBERTa) in Sect. 2.3. A literature review follows, summarizing fake news (Sect. 2.4) and hate speech detection (Sect. 2.5) methods, including lexicon-based, machine learning, and deep learning approaches. The chapter also introduces related work that influenced the experimental design in subsequent chapters in Sect. 2.6. Finally, it summarizes this chapter in Sect. 2.7.