(Note: This topic was already discussed with the dev and another Korean contributor on Discord. I'm just making this as a public post so those who haven't joined Discord yet can also be informed.)
I found that the biz.chosun.com domain is already marked as AI in search results. That website is from a major newspaper in South Korea. My guess is that since they're translating Korean articles into English via AI without proper proofreading (which is definitely an issue), people who don't know much about Korea might have reported it as AI-generated.
One issue is that because the AI label is applied at the domain level, it affects all content from that domain. Even the original Korean articles are being marked as AI (see the screenshot below), so I think the labeling needs to work at a more granular level.
I understand that Kagi Translate (which is also LLM-based) is preparing a business API. This also raises a concern about whether all AI translations should be classified the same way as low-quality pages.
Therefore, I suggest creating a separate label specifically for "AI-translated" content to distinguish it from AI-generated spam.
