I see you edited and so I thought I'd follow up with some thoughts.
It isn't clear where AI queries are processed. They use to have an infobox saying what LLM model it was and that it ran on their systems, but that seems to be gone. They have also since released an AI assistant product that connects to many providers so I'm guessing your searches with question marks are proxied to other companies. Note also that for normal (non-ai) searches, Kagi uses google and other indexes to provide results, so your search is also going to them. But it's proxied so google doesn't know which is your search term and which is mine.
Virtually every major privacy focused search engine does this, as running a full search index is incredibly expensive, so outside of google and bing there aren't many good ones that run their own index and never provide results from other search engines.
Also, is there anything bad about them, like affiliated with Google, satan, Facebook, or some other undesirable entities.
They are a start up, have no VC funding according to other lemmy people, and since you're paying for the service (instead of advertising being their revenue) they are incentivised to make the best search instead of show you more ads.
Basically, we don't know if they are evil but as far as we know they aren't. Not owned by google, Facebook, etc but I don't know if satan owns them.