Large NLP machine learning models and their upcoming impact on cyber security - Geoff McDonald

Recent advances in large natural language processing (NLP) models are starting to change the threat landscape in cybersecurity. In this talk, I'll begin by introducing the latest NLP innovations focusing on large GPT generative models, and basic related concepts including zero shot learning, few shot learning, and fine tuning. I'll then demonstrate a bunch of immediate risks to cybersecurity from these models - including automated targeted phishing email generation, automated initial access loaders, and disinformation generation. For this talk you do not need any background in machine learning or NLP, it will be a beginner-friendly talk.