Chinese and western scientists identify ‘red lines’ on AI risks - eviltoast