Getting from Generative AI to Trustworthy AI - arxiv.org