nobodywasishere You previously said that 98% of reports were accurately identified as AI generated, and even if we ignore the portion of that remaining 2% you dismissed as not AI, this means Kagi has a user base capable of detecting AI content with 98% accuracy.
98%!
Yet apparently your intention is to keep this high-quality contributor base's input for yourselves while building your own commercial models.
Nobody argued that we should decide AI content based solely on vote counts. What we said is that those who flag suspicious sites should initiate transparent, public discussions — not private reports. If someone else's report changes my search results, I absolutely need to know why. Otherwise, you contradict your own stated mission: "Our mission is to humanize the web."
I understand your fear of Google or Bing using this dataset. But you could at least consider a query interface exclusive to Kagi Subscribers, or another alternative solution.
Please don't get swept up in the AI hype while ignoring your users.
