How do you see LLMs handling the constant flow of new data, given spam, SEO hacking, fake news and what not? Can the training be reliably automated and performe

How do you see LLMs handling the constant flow of new data, given spam, SEO hacking, fake news and what not? Can the training be reliably automated and performed that quickly? It seems the chatGPT folks are allowing free access so they can test and fix abuses with human intervention. But that's hard to do in real time.