Novo Nordisk Breach Exposes AI Models and Patient Data

Novo Nordisk confirmed a cybersecurity breach (June 11–12, 2026) in which attackers exfiltrated extensive proprietary AI assets — including a 16.7GB multimodal model (Dragonfly), training datasets, full source code, 113 training-run logs, container images (~53GB), infrastructure maps, and pseudonymized clinical-trial patient data — and have published evidence while attempting to extort the company; external investigators and regulators have been notified.