Any resources for adversarial anti-AI countermeasures?

I'm looking for links and reading material about flaws of the current generation of AI bots/summarizers, that one can exploit and insert into their websites, art, or anything that is currently used as training model, or processed by AI, to create bogus or useless results.

I am aware of the concept of prompt injection, I'm interested in practical examples of it or other grey hat tools one can leverage to disrupt LLM/ChatGPT & co.

Any pointers?