I'll answer the questions in reverse order because the policy defines whether a banner is needed.
Policy
Currently there's no local code golf consensus on ai answers - a search for "ai" on code golf meta reveals no policy about banning usage of chatgpt and co. Further, there is no official network wide policy disallowing referenced ai posts:
With due consideration, we've decided no network-wide, general policy regarding banning ChatGPT, or other AI generated content, is necessary or helpful at this time.
Quote from an answer by Community Manager (CM) Slate regarding banning chatgpt network wide.
Noticed how I said that referenced ai posts are not disallowed (that is, ai generated posts that are properly cited as being made by ai are allowed by default). Posts suspected of being generated by AI without proper referencing are plagiarism and aren't allowed regardless of ai policy.
Therefore, the question becomes whether properly referenced ai generated posts are allowed. Personally, I'd lean towards allowing them because it's gosh dang impressive if you can get chatgpt to create a golfed answer that actually solves a challenge and abides by the competitive submission policy. Trust me I've tried from my experience it makes code longer.
Banner
Therefore, given the lack of policy, a banner isn't needed at this stage. How can people be warned about something that doesn't exist? If at some point there's a policy decision that ai generated posts that are properly referenced aren't allowed, then the banner would be a good idea.