In Canada there was a company using an LLM chatbot who had to uphold a claim the bot had made to one of their customers. So there's precedence for forcing companies to take responsibility for what their LLMs says (at least if they're presenting it as trustworthy and representative)