Prompt-based data generation using generative models can be used for a wide range of defensive actions by assisting, and reducing the workload of, security analysts, and enabling partial automation. Processing cybersecurity logs and analyzing complex data distributions for anomaly detection, performing data augmentation, attack simulation, and feature extraction for intrusion detection and malware classification are just some of the application areas. This chapter discusses the use of generative pre-training, generative adversarial networks, and variational autoencoders in cybersecurity.