Why AI safety is like a bolt in a croissant - Riskgaming | Wave AI Podcast Notes