If you pay someone for a number of posts, you’re paying for outcome. Unless you condition the outcome with “no AI”, it’s fair game.
Whether that’s ethical without the condition is a different question: the implication you could draw is “I’m paying for posts to encourage engagement, they should be honest and human“ but unless you make it clear this is the intent (to imply the outcome you want), it’s *fine*.
In the same way, if you pay for “20 posts”, the person could reply to 20 topics with the word “sausages” and that would be *fine* in theory. Because unless otherwise specified, you haven’t qualified what the 20 posts should contain, what is acceptable, what is not.
Whether AI is permitted or not is no different. If you then say “no AI” and they sneak it past you, that *is* unethical and not in accordance with the terms of the agreement made - if you can prove AI was used against your instruction…