Reasonably foreseeable misuse is the use of an AI system outside its intended purpose in a way that can still be anticipated from human behavior or interaction with other systems. In governance and regulatory contexts, it requires providers to consider misuse scenarios that are plausible, not merely intended uses.