The new UK government's stance on AI safety — LessWrong