Bounty spammer meets AI Chat generator. Can they be banned on the spot? - page 5.

joker_josue

legendary

Activity: 1722

Merit: 4711

**In BTC since 2013**

Quote from: LoyceMobile on January 19, 2023, 07:02:09 AM

I found another user that looks like a generic chat generator spammer:
https://bitcointalksearch.org/user/cryptounitycut-3524806

I would also say that these texts are generated by the AI. Because, instead of answering the question directly, he makes a little explanation and then he answers the question in a very comprehensive way.

LoyceV

legendary

Activity: 3290

Merit: 16489

Thick-Skinned Gang Leader and Golden Feather 2021

Quote from: dkbit98 on January 19, 2023, 04:34:54 PM

Quote from: LoyceV on January 18, 2023, 01:47:03 PM

I tried that in another topic. Conclusion: I'm a bot Tongue

Since I'm obviously human, the detection is unreliable.

You didn't say what was the percentage showing text was AI generated on all 3 websites...

The site I tested before labeled me as 81% fake.

Quote

How do you think we are detecting plagiarized text by other members?
We certainly don't have super powers, and I was always using two website to detect source and PERCENTAGE of plagiarism... it worked well most of the times.

I always use an exact match on them, or manually find word spinning. I've never used a "plagiarism detector" for it.

dkbit98

legendary

Activity: 2212

Merit: 7064

Quote from: LoyceV on January 18, 2023, 01:47:03 PM

I tried that in another topic. Conclusion: I'm a bot Tongue

Since I'm obviously human, the detection is unreliable.

You didn't say what was the percentage showing text was AI generated on all 3 websites... but some people have been saying you are a bot for years Cheesy

Quote from: LoyceV on January 18, 2023, 01:47:03 PM

Based on a third-party tool? I doubt it.

How do you think we are detecting plagiarized text by other members?
We certainly don't have super powers, and I was always using two website to detect source and PERCENTAGE of plagiarism... it worked well most of the times.

EDIT:
Here is one test I did with this post from member Techdev generated by AI (it was later deleted by moderators after my report):
https://ninjastic.space/post/61619530

GPT-2 Output Detector is showing 99.98% that this is AI generated:

GPTZero is saying that text is most likely to be AI generated:

LoyceMobile

hero member

Activity: 1659

Merit: 687

LoyceV on the road. Or couch.

I found another user that looks like a generic chat generator spammer:
https://bitcointalksearch.org/user/cryptounitycut-3524806

Typically, it adds as much words or phrases as possible to make the post or message sound longer and better. << See what i did there?

It's more annoying that normal shitposts because it takes more time to notice.

LoyceV

legendary

Activity: 3290

Merit: 16489

Thick-Skinned Gang Leader and Golden Feather 2021

Quote from: dkbit98 on January 18, 2023, 07:24:48 AM

It gives you percentage that shows if some text was potentially generated by AI bot or not.

I tried that in another topic. Conclusion: I'm a bot Tongue

Since I'm obviously human, the detection is unreliable.

Quote

If one member gets repeated high percentage of AI generated posts I think this member should be banned or temporary banned.

Based on a third-party tool? I doubt it.

dkbit98

legendary

Activity: 2212

Merit: 7064

Quote from: LoyceV on January 16, 2023, 04:16:53 PM

Since it can have both false positives as well as false negatives, I don't think it's worth the effort.

It gives you percentage that shows if some text was potentially generated by AI bot or not.
We use similar thing to detect if something was plagiarized in forum and report that to moderators, so I think we can use something similar with AI generated text.
If one member gets repeated high percentage of AI generated posts I think this member should be banned or temporary banned.

Quote from: LoyceV on January 16, 2023, 04:16:53 PM

Also: what are the odds those "detectors" only use the current AI chat for their own publicity?

I don't know who owns those websites I posted and I don't care who they are, except one that is collaboration between people from MIT and Harvard.

Ultegra134

hero member

Activity: 1680

Merit: 845

Quote from: DaveF on January 15, 2023, 12:04:33 AM

Quote from: Ultegra134 on January 14, 2023, 03:54:35 PM

Quote from: LoyceV on January 14, 2023, 02:47:21 AM

My guess: he's trying to earn that one Merit needed for a Jr. Member signature for bounty spam.

Wasn't aware that there are bounty campaigns in the altcoin section have such low requirements.

Perhaps we are looking at it the wrong way. Ban the campaign managers and the projects that hire them if they use people who are using spam bots and give everyone in the campaign a warning that they were working with spammers. After a certain number of warnings you would get a ban. Would that be too harsh?

Seriously, I use the altcoin section for a few things and would not be put at all if it went away, so it might be a harsh point of view, but to me it looks like most of the bounty campaigns are a lot crappier then they used to be.

-Dave

Well, it does sound a little harsh, but it depends on how you see things. In general, the most spam and gibberish posts stem from the 1xbit signature campaign (apart from the altcoin section, which you mentioned), and despite the efforts of plenty of users who reported a large number of their posts, the issue remains. Two sections that are filled to the brim with such posts are the gambling and altcoin sections, in which I hardly ever participate; thus, I can't have a comprehensive reason for the spam, but in my opinion, the gambling requirements that campaign managers put in are one of the possible causes. The altcoin section, on the other hand, is filled with bounty campaigns that are attracting newer users or those who are unable to enter the most reputed ones.

Ryu_Ar1

hero member

Activity: 910

Merit: 677

Quote from: LoyceV on January 12, 2023, 01:15:15 PM

If you find more of this, please post them here as examples.

Sorry in advance, are posts like this really allowed?

I don't really understand the language the OP uses, but if you look at the post, most of it is like that, even though there are a few quotes there as well as a few pictures, but looking at the whole it looks like you will understand what I mean.
https://bitcointalksearch.org/user/karabiber-556829

Sandra_hakeem

hero member

Activity: 798

Merit: 1045

Goodnight, ohh Leo!!! 🦅

Quote from: LoyceV on January 16, 2023, 04:16:53 PM

Quote from: dkbit98 on January 16, 2023, 12:44:17 PM

@LoyceV can you check and test this tools I found for detecting AI bots?

Since it can have both false positives as well as false negatives, I don't think it's worth the effort.
Also: what are the odds those "detectors" only use the current AI chat for their own publicity?

I haven't really been been cleared on that Yunno; cus I see -- getting publicized -- as one huge reason why they go about with artificial intelligence in the first place...... ofcourse,I don't think if they wanted to keep this nuisance a secret, because what's the point of having DETECTORS?? Maybe it was made to gain some top-notch publicity as you said ( though I think less of that) or some peeps, after getting tired of the whole drama, had to get into their worst primes just to get 'em either granted unopposed as something very important, or segregated as something unnecessary at all.
The latter seems true to me .

Sandra 🧑‍🦰

LoyceV

legendary

Activity: 3290

Merit: 16489

Thick-Skinned Gang Leader and Golden Feather 2021

Quote from: dkbit98 on January 16, 2023, 12:44:17 PM

@LoyceV can you check and test this tools I found for detecting AI bots?

Since it can have both false positives as well as false negatives, I don't think it's worth the effort.
Also: what are the odds those "detectors" only use the current AI chat for their own publicity?

dkbit98

legendary

Activity: 2212

Merit: 7064

Quote from: DaveF on January 13, 2023, 12:37:26 PM

So...
NotATether, ETFbitcoin, o_e_l_e_o, LoyceV are all against my concept of a captcha.

I am also against adding captchas for writing every post, and it would create big decrease in forum activity.
Maybe adding captchas could be considered for people who are writing many worthless posts in short amount of time.

@LoyceV can you check and test this tools I found for detecting AI bots?
This tools can be used freely and it's called GPT-2 Output Detector, GLTR and GPTZero.
They are not 100% correct all the time, but they can be indicator if someone is using AI for making posts.

Sources:
https://openai-openai-detector.hf.space/
http://gltr.io/dist/index.html
https://etedward-gptzero-main-zqgfwb.streamlit.app/

Ucy

sr. member

Activity: 2674

Merit: 403

Compare rates on different exchanges & swap.

In certain gatherings comments like that would be given massive round of applause not because they make any sense, are relevant or true, but because of the "powerful" grammer or vocabularies that are used.
It would be great if people focused more on relevant comments (comments relevant to what is being discussed) even if their grammar or vocabularies seem poor/weak. I have observed certain posters who don't really write well but have strong points and relevant comments. You would notice their posts value or relevance if they were rewritten by a qualified English professor. And you could easily notice the meaninglessness of AI generated texts for example if they were rewritten by someone with poor grammar.

I believe it's best to attract those with strong interest & passion for Bitcoin/crypto otherwise we will have a forum full of people applauding or even meriting meaningless comments/posts because of their powerful grammar. Such comments will likely be boring and easily noticed by passionate members even though they look attractive. And the passion could help in the understanding of relevant/reasonable comments written by members with poor English

o_e_l_e_o

legendary

Activity: 2268

Merit: 18711

Quote from: DaveF on January 15, 2023, 12:04:33 AM

Perhaps we are looking at it the wrong way. Ban the campaign managers and the projects that hire them if they use people who are using spam bots and give everyone in the campaign a warning that they were working with spammers. After a certain number of warnings you would get a ban. Would that be too harsh?

Nope, not too harsh at all. In fact, doing exactly this is already in the rules stickied to the top of the Services board:

Quote from: hilariousandco on November 17, 2016, 07:46:01 AM

If you are running a campaign and it becomes blatantly obvious to Staff that you are doing little to nothing to stop spam on your campaign you will be issued a PM warning by a Global Moderator that you need to make immediate improvements to curb low-quality posts. You will have 7 days to remove low-quality posters and respond to the message detailing what you are going to do to make changes to your campaign to reduce the amount of spam. If improvements are not noticeable within 21 days of that and Staff do not believe you are doing enough to prevent low quality posts your signatures will be blacklisted from the forum by an Admin and you will no longer be permitted to advertise here in such a way.

As far as I am aware, this has literally never happened. Even when we had the absolute flood of YoBit spammers, it took months of the forum being an absolute shitshow and multiple threads in Meta and Reputation complaining about it before we finally got some intervention.

Just enforce the rules. 7/14/30 day escalating bans for 1st/2nd/3rd offenses for spammers, followed by a permaban. Same for campaign managers and campaigns themselves. Actually do this and the problem is solved. Spammers are banned, and new spammers joining are massively reduced because there is no incentive for them to do so, since any campaign manager which accepts them will also be banned. Campaigns are incentivized to use good campaign managers, because if they choose a bad one they will also be banned from advertising here.

The solution already exists. It's just the mods and admins ignore it. I don't know why.

DaveF

legendary

Activity: 3500

Merit: 6320

Crypto Swap Exchange

Quote from: Ultegra134 on January 14, 2023, 03:54:35 PM

Quote from: LoyceV on January 14, 2023, 02:47:21 AM

My guess: he's trying to earn that one Merit needed for a Jr. Member signature for bounty spam.

Wasn't aware that there are bounty campaigns in the altcoin section have such low requirements.

Perhaps we are looking at it the wrong way. Ban the campaign managers and the projects that hire them if they use people who are using spam bots and give everyone in the campaign a warning that they were working with spammers. After a certain number of warnings you would get a ban. Would that be too harsh?

Seriously, I use the altcoin section for a few things and would not be put at all if it went away, so it might be a harsh point of view, but to me it looks like most of the bounty campaigns are a lot crappier then they used to be.

-Dave

Ultegra134

hero member

Activity: 1680

Merit: 845

Quote from: LoyceV on January 14, 2023, 02:47:21 AM

My guess: he's trying to earn that one Merit needed for a Jr. Member signature for bounty spam.

Wasn't aware that there are bounty campaigns in the altcoin section have such low requirements.

Quote from: ?? on ??

CAPTCHA can block fully automated bot which automatically select thread at random, obtain text from AI (such as chatGPT) and then make a reply using that text. But it's unlikely it'll stop people who copy text manually from website which provide AI service.

There probably is a small number of users who might be using such a bot, but in my opinion, as you've already mentioned, the majority will just stick to manual copy and pasting from AI services. Using a fully automated bot requires some knowledge that the majority of these spammers don't have, while it sounds way too exaggerated. The manual method is far easier and more hassle free.

o_e_l_e_o

legendary

Activity: 2268

Merit: 18711

Quote from: LoyceV on January 14, 2023, 02:47:21 AM

Combining those 2, I can think of a legit reason for word spinning: to avoid identification by using forensic linguistics. But that's not the case in any of the spam-cases we've seen.

Word spinning does a very poor job at this, since you keep the same sentence structure, grammar, etc. But still, none of the spamming accounts using word spinners are word spinning some text that they wrote themselves. They are plagiarizing.

Quote from: ?? on ??

CAPTCHA can block fully automated bot which automatically select thread at random, obtain text from AI (such as chatGPT) and then make a reply using that text.

But equally it is trivial for the bot to do all this anyway and only require human interaction for <10 seconds to solve the captcha. If you are not using Tor or something else that means you end up having to solve the same captcha twenty times or more, then one human can still mass spam with multiple bot accounts with little hindrance.

Just ban the spammers.

Adbitco

hero member

Activity: 1428

Merit: 653

Leading Crypto Sports Betting & Casino Platform

Quote from: DaveF on January 12, 2023, 10:07:59 PM

I wonder how difficult it really would be to add a captcha requirement for every post. I know some boards have them, sometimes even multiple ones. I don't know if it hurts legitimate posting or not, but I don't see as much spam on them vs those that do not. One car board I am on has a quick and simple one like what is 2 + 7 and that is followed by the 'pick the image' one. And there is very little spam. The other is more like here where you can just post and does get a lot more junk posts.

Don't know how much work it would take to put one in.

But, the AI / chat spammers should just get a permanent vacation.

-Dave

Well, sorry commenting here, i think your opinion could be the best.
Sometimes i think of how much we could keep reporting such post or even issuing temp ban and parma ban., However, have we forgotten that the more they got ban the more account they make and if their IP are already mark for evil, believe me or not they could source for another means to create additional account to still further same process, will the admin keep banning them on the go?
In as much as we want to keep the forum clean we should also try to lessen the task from the admin & Mods, because i believe there are lot cases to be taking care of over here. For such to reduce they could implements a captcha to reduce much activities from them.

LoyceV

legendary

Activity: 3290

Merit: 16489

Thick-Skinned Gang Leader and Golden Feather 2021

Quote from: o_e_l_e_o on January 13, 2023, 03:14:21 PM

controversial opinions
~
Word spinning should be an instant perma-ban. I cannot fathom why it isn't.

Combining those 2, I can think of a legit reason for word spinning: to avoid identification by using forensic linguistics. But that's not the case in any of the spam-cases we've seen.

Quote from: Ultegra134 on January 13, 2023, 03:33:19 PM

I believe that the user in question is likely to be an alt account; I can't find any other reason a newbie would make such a large amount of posts in such a short notice.

My guess: he's trying to earn that one Merit needed for a Jr. Member signature for bounty spam.

Ultegra134

hero member

Activity: 1680

Merit: 845

Quote from: DaveF on January 13, 2023, 12:37:26 PM

So...
NotATether, ETFbitcoin, o_e_l_e_o, LoyceV are all against my concept of a captcha.

Let's compromise and do it my way :-)

Seriously, it would be a patch at best to slow it down but other then reporting we have no other option.

IMO it's just a really difficult balance, you want a free and open forum. But you want to control the spammers as much as possible.

Or do you go the other way but no more temp bans for things like that. Using a word spinner you get the boot and that's it. Too harsh? Not sure.

-Dave

I'm sorry, I don't mean to offend, but how are captchas in each post going to help? In the case of LoyceV's report, the reported user has posted four replies in a course of approximately 40 minutes. These posts are certainly made by an AI service such as ChatGPT, as LoyceV already mentioned. I don't see how a captcha would solve that; I don't see the need, since these kinds of services are able to generate a vast amount of posts.

Anyway, to return to the starting subject, I believe that the user in question is likely to be an alt account; I can't find any other reason a newbie would make such a large amount of posts in such a short notice.

o_e_l_e_o

legendary

Activity: 2268

Merit: 18711

Quote from: DaveF on January 13, 2023, 12:37:26 PM

Seriously, it would be a patch at best to slow it down but other then reporting we have no other option.

The other option is mods and admins actually enforce the few rules we do have. How much better would the forum be if we actually followed this in any way: Signature Campaign Guidelines (read this before starting or joining a campaign)

Quote from: DaveF on January 13, 2023, 12:37:26 PM

IMO it's just a really difficult balance, you want a free and open forum. But you want to control the spammers as much as possible.

There is a very clear distinction between allowing freedom of speech for controversial opinions (which I will always defend) and banning outright spammers.

Quote from: DaveF on January 13, 2023, 12:37:26 PM

Using a word spinner you get the boot and that's it. Too harsh? Not sure.

Using a word spinner is plagiarism. Additionally, using one is deliberately trying to circumvent the rule regarding plagiarism. Word spinning should be an instant perma-ban. I cannot fathom why it isn't.

Topic: Bounty spammer meets AI Chat generator. Can they be banned on the spot? - page 5. (Read 2557 times)