To date, at least 1,200 reports of AI incidents have been recorded in various public and research databases. That means that now is the time to start planning for AI incident response, or how organizations react when things go wrong with their AI systems. While incident response is a field that’s well developed in the traditional cybersecurity world, it has no clear analogue in the world of AI. What is an incident when it comes to an AI system? When does AI create liability that organizations need to respo