Who do we hold accountable for rogue AI? The companies? The user? Surely not the victim or those affected by AI's output. I might come back to this and flesh this out a bit more, but for now it's a question I continue to come back to.