AI safety standards

New UN-backed assessment warns that weak oversight could leave societies exposed to cyber, biosecurity, disinformation and control risks.