As digital assessment continues to evolve, concerns around collusion, AI-assisted misconduct, and academic dishonesty remain top of mind for assessment leaders. This session will present an exploration of practical approaches to collusion detection, framed within the broader concept of assessment security. Drawing on real-world discussions and applied examples, this session examines how multiple signals can be used to flag potentially problematic behaviour. Topics include text similarity metrics, identifying actions such as copy-and-paste and large text insertions, and emerging techniques like the use of invisible text embedded within prompts. It also considers how AI both contributes to collusion risks and how AI-informed techniques are being explored to detect and prevent collusion. Participants will leave with a clearer understanding of how layered, data-informed strategies can support fair, secure, and defensible assessment practices.