Explore how Long Short-Term Memory (LSTM) recurrent neural networks combined with Python can leverage Artificial Intelligence for generating invalid, unexpected, or random data inputs in software testing and network security vulnerability detection. Delve into the concept of fuzzing, an established testing approach using machine-generated test inputs, and learn how various types of machine learning can produce novel outlier test cases superior to purely random inputs. Discover the process of training neural networks to create more effective fuzzing data, building upon work with image processing using support vector machines (SVM) and Generative Adversarial Networks (GAN). This beginner-level talk, presented at EuroPython 2018, aims to provide software developers and testers with a fundamental understanding of fuzz testing, covering topics such as neural fuzzing basics, fuzzing mechanics, steps for training deep neural networks, and implementing a basic LSTM neural network for security testing.