How soon before we see the results of spam influencing LLMs and the output of AI?

It's quite possible to influence this in a big way, and I wonder what steps the LLM owners are taking to stop this becoming the next problem.

I'm sure I haven't given any ideas away.