Presentation
Digital twins are moving from visualisation tools to decision-making infrastructure. Predictive maintenance, remote survey, and voyage optimisation increasingly act on the outputs of AI-enabled models - creating a continuous data pipeline spanning vessel, shore, vendor, and cloud, and with it, an exposure the sector's vessel-level compliance mindset was never designed to defend.
This session examines what it takes to trust AI-enabled twins in a contested environment. It sets out the principal failure modes - compromised data provenance, model drift and manipulation, the OT-to-cloud pipeline as a target, and automation bias in those acting on the outputs - and their consequences for operational and safety-critical decisions. It then offers a practical assurance approach grounded in current class and international frameworks, and considers the emerging link between demonstrable twin assurance and the insurability of AI-informed operations.