Quote from: Ultegra134 on Today at 06:36:07 PM

Quote from: memehunter on Today at 10:56:20 AM

I have a query for @Ultegra134, @nutildah, @lovesmayfamilis, @Dark.Look, + any other member who deals with AI detection.
I checked the post history of one account (can't disclose, don't want to make the AI spammers smarter + I can't prove anything right now). Out of his last 50 posts, 32 are showing 100% detection by Sapling AI. No other detector like Copyleaks and GPTzero is showing AI. Posts are long and detailed (IMO, AI). What do you make of it?
IMO, this is a case of humanizing AI content. I mean, is it normal to have such a high amount of false positives by Sapling AI? Have you encountered this before?
I can pm the details to @Ultegra134, @nutildah, @lovesmayfamilis with the condition of confidentiality.
I am not accusing anyone right now. It is a part of my brainstorming to detect AI spammers who are using humanizers.

Feel free to contact me with more details. I've had a few false positives (or negatives) with Sapling.ai, but never that many as in your case. I don't consider it a 100% trusted AI detector, but it hasn't failed me in most cases. Let me know which user it is and I'll check their posts as well, if something is fishy I'll get back to you as soon as possible. What other detectors did you try out, except GPTZero and Copyleaks? Originality and Winston.ai are 95% spot-on, but they're paid tools and if my memory serves me right, Originality stopped providing free credit trials.

I have checked in Copyleaks and Origin GPTzero only apart from Sapling AI. I know Sapling AI can give false positives, but IME, this is the only one that is hard to bypass if you are using humanizers or copy-paste translation.
I have pmed you the user's profile in question.
Yes, many good AI detectors now require paid subscriptions. You can bypass that with vpn and all, but I am not suggesting that Wink