The rational design of proteins and peptides is a longstanding challenge in molecular biology, with broad implications for biotechnology, therapeutics, and synthetic biology. Traditional methods—such as directed evolution and structure-guided engineering—are limited in scope and throughput, often failing to efficiently explore the vast combinatorial space of amino acid sequences. This work aims to comprehensively review and analyze the application of generative artificial intelligence (AI) models for the de novo design of proteins and peptides, highlighting recent advancements, dataset resources, generative methods, and their real-world implications. We examine state-of-the-art generative architectures, including Generative Adversarial Networks (GANs), Transformer-based language models, diffusion models, and reinforcement learning strategies. These methods are assessed based on their ability to model protein sequence-function relationships, generate novel sequences, and optimize specific biochemical properties. We also discuss the role of curated datasets (e.g., UniProt, APD3, PDB). Generative AI models demonstrate substantial capabilities in producing novel, diverse, and functional proteins and peptides with application-specific properties. Several studies show AI-generated candidates outperforming traditionally designed ones in stability, activity, and novelty. Integration with downstream pipelines—such as structure prediction (e.g., AlphaFold2) and in silico docking—further enhances their utility. Generative AI offers a paradigm shift in protein and peptide design, enabling rapid and scalable exploration of biomolecular sequence space. Its applications span industrial enzyme optimization, therapeutic peptide development, synthetic circuit design, and personalized medicine. As experimental validation pipelines mature, these models are poised to become foundational tools in computational biology and bioengineering.