As the AI language models are getting sophisticated, so are the social engineering attacks. In this session, we look at how attackers leverage AI in creating convincing phishing emails, voice deepfakes, and chatbot scams. We will analyze real-world examples and go into the technical patterns behind AI-generated attacks, developing practical strategies for detection. Learn how large language models are misused for scams, how to track their telltale signs, and what tools and techniques will help you build more effective defenses against them.

In this, we will also take a look at the current landscape of AI-powered social engineering and recent attacks that have successfully breached organizations. Learners will discover the most important indicators distinguishing AI-generated content from human-written text: linguistic patterns, contextual inconsistencies, and metadata analysis. The session will show how multiple AI technologies—from text generation to voice synthesis—are combined by attackers to create multi-channel attacks, which are increasingly difficult to detect.

Most of the talk will focus on detection and defensive strategies. In much detail, we will go over common patterns in AI-generated phishing attempts, including how language models can be elicited to create contextually aware scam messages. Students will learn about the limitations of current AI models and how these limitations can be leveraged for detection. We will walk through a number of available tools for identifying AI-generated content, starting from simple pattern matching to more sophisticated probabilistic approaches.