I don't see much point in that from a practical standpoint, you don't really need a LLM to generate spam, and content is not the only way spam is detected. But

I don't see much point in that from a practical standpoint, you don't really need a LLM to generate spam, and content is not the only way spam is detected. But it may happen just because they can. Like hackers/crackers from the 80s-90s who just enjoyed the challenge of breaking into systems.