x
What AI Governance and Policy Can Learn from "AI Models are Capable of In-Context Scheming”? — LessWrong