AI debates its own ethics at Oxford University, concludes the only way to be safe is “no AI at all”