A two-peer analytical conversation on why AI incidents should be treated as organisational events rather than purely technical failures. The discussion moves from defining what counts as an AI incident, to governance and decision rights, to the first 24 hours, then investigation, communication, learning and rehearsal. The exchange keeps legal points at general-information level and focuses on practical leadership implications grounded in the source.