The potential risks of AI are so macro and meta that I'm not really sure reporting standards or security assessments are gonna do anything to address the true risks.