bioweapon

All highly capable AI models should be regulated through pre-deployment and ongoing testing. But any safety regime that works for domestic models should work for Chinese ones.