Thank you! My concerns are with all models, including the LLMs. Partly because we do not understand how they learn, and partly because they are so powerful. I am speculating in the article about their potential negative uses. Just like a hammer that can be used for building things or destroying things.