AI Sites - Cover Your Tracks Or 'Nah'?

[Thonglor]

Hey all - White hat and BUSO newbie here.

I'm contemplating using AI assistance to start up a few sites as a proof of concept for a few niche ideas. 'AI-assistance' being a combo of human researched content and AI written text. Obviously, no ground-breaking ideas here. I've seen many niche site owners here and in other forums doing this.

The question is, are you going at length to keep your AI sites separate from white hat ones? Different Google accounts, servers etc... to maintain separation between them. Recent updates especially the HCU has me concerned about the potential effect on non-AI ones.

Over-thinking? Or warranted? If you're covering your tracks, what's your setup like?

 

[Elvis]

There are a few situations where I would look to cover my tracks. Namely, when there are likely to be legal consequences. (I'm not referring to anything that's specifically illegal; rather, things that could to annoying and unnecessary lawsuits, as happens here in the USA.) Working with AI to write content would not fall under this umbrella in my opinion.

Major sites have been using AI to write content for years. Recently, Google put out a new article and/or updated content guidelines to specify that the use of AI for content generation is not inherently bad. I could certainly be missing something, but even if you overdo it, I don't see this as a violation that warrants Google tracking down everything you're connected to and issuing a manual action.

Another way to look at it is, what is the goal for these sites? Do you want to run display ads or affiliate links? If so, you would need more than a separate Google account and hosting server. You'd need separate business entities to then get separate accounts with Google Adsense + AdThrive, Mediavine, etc. Or else you have footprints.

 

[Thonglor]

Thanks, Elvis. The goal is pretty straightforward - ad and affiliate revenue. Even though it would be checked for accuracy and sound good, it won't ace any classifiers either.

And you're right, you can't escape entering your AdSense ID for MV, AT or most networks for that matter.

 

[alexsimon]

No, I do not cover tracks but I never publish AI-generated content without a thorough review. And modifications, so many in fact that they pass Ai verification tools most often. Also, I will never have 100% AI-generated sites. Some important pages will always be manually written, sales pages, homepages etc. I don't think there is a need to cover any tracks as it is not black hat.

 

[Kakeshi]

I feel for money sites with big earning potential, covering the tracks is necessary by ensuring the quality of the AI content. As search engines are warming up to face the sudden rise in AI generated content, expect some changes in the algo in the days to come.

 

[Thonglor]

No, I do not cover tracks but I never publish AI-generated content without a thorough review. And modifications, so many in fact that they pass Ai verification tools most often.

Thanks Alex. Is there a specific verification tool you're using? I've tested using ZeroGPT and Open AI's own classifier. I found ZeroGPT the most accurate.

How much time do you take to fine-tune your articles? I'm using these sites as a proof of concept so scaling quickly is a necessity. But of course, not to point where it's inaccurate and a steaming pile of ...

I feel for money sites with big earning potential, covering the tracks is necessary by ensuring the quality of the AI content. As search engines are warming up to face the sudden rise in AI generated content, expect some changes in the algo in the days to come.

Yes, my paranoia thinks the same but I'm trying to hear all sides.

Don't mind me asking, how are you keeping them separated? Especially @Elvis comment about AdSense? Unlike the early days, those accounts are hard to come by.

 

[alexsimon]

I use originality.ai and to a lesser extent huggingface's gpt checker (it was used for gpt-2 but still works fine and it is free)

 

[cepen]

I'm not really separating sites made with AI, at least not at the server level.

I create new accounts for Analytics, Search Console, I don't link between sites... but in the end I plan to put Adsense or Ezoic from the same account so I'm not too afraid.

I'm just trying to keep the AI generated content from being junk. I think that Google in the medium and long term will not be able to fight this type of content, as long as they are well done and provide value to the user.

I hope I'm not wrong and that all my projects go to zero haha.

 

[Ryuzaki]

I think Google has been very clear about one thing and has tried to insinuate another.

The thing they're clear about is AI content is okay, you won't get penalized. It just needs to be for the reader first and needs to be high quality, as it will be judged in that fashion like any other content.

What they're saying between the lines if you read all the documentation is that there is an expectation that you ADMIT that your content is written by AI.

Here is where my opinion about the future of this comes in:

They'll know it is, and if you don't admit to it, that's going to be like a zero multiplier against your score, or some really low number. You'll have an anchor tied to your legs and won't rank well no matter how good the content is. Why? Because of EEAT and the expectation that there is an author. And if you claim "Johnny Two Shoes" is the author when Google knows it's AI, then that's zero EEAT.

But if you say "this content was produced by AI and fact checked, reviewed, edited, and punched up by Johnny" then you won't get a zero multiplier. Because you've told the users what they need to know, as in WHO wrote the content. My opinion is, if you're going to do AI content, this is the way to do it.

My other opinion is you'll still get a "less-than-one" multiplier on your overall ranking score, meaning you don't have an anchor tied to your legs, but maybe a few bricks as you swim up towards the surface of the SERPs. Why? Because EEAT, which is the solution to all this AI content. An AI can only take old content it was trained on and copy it, it can't bring any new information to the table. I can't think of the technical term but there's no new "information load or expansion or addition". It's the same old crap.

And it's not written by anyone this time. How much credit should a team of doctors get if they just copy and pasted information from other sites and signed off on it saying "okay this is accurate". They should get zero credit and it should go to the original sites. That's the conundrum.

So anyways, do you need to cover your tracks as in hide footprints and not let Google know you own more than one of these sites or whatever? No, they've explicitly said it's okay. That doesn't mean you'll rank, though.

Also you won't hide it from them. They won't find you algorithmically or care, but if they wanted to catch you they would. Their cookies are everywhere, you're probably loading Google Fonts on your sites, using Sheets or Docs, Gmail, Adsense, Double Click for Publishers is being used by everyone ad network, so on and so forth. They do go to these lengths for very vindictive manual penalties (I've had one), but they won't do that for AI sites. But my point is if they want you, they've got you already. お前 和 網 死んでいる.

 

[cepen]

100% agree with Ryuzaki. I agree that if they want to find you, they will find you.

But I reflect and go a bit further, more long term.

The ways in which we users are using AI are still very simple, even the AI services themselves are very new (just a few months ago we didn't have ChatGPT).

I believe that not too long from now AIs will be able to generate valuable content, not only with old content.

If they connect to the Internet and from the information we give them, they will generate new ideas, conclusions, etc...

If we, the users, pass raw (new) data to the AI so that it can generate new ideas, texts, conclusions, the game changes. A simple example: we give the AI hundreds of real opinions of users who have bought a certain product and we ask it to generate an opinion or a review based on this information.

Another example: a journalist gives the AI the highlights of an event or an investigation and asks the AI to help write a news story that is as easy to understand as possible.

The possibilities are endless. It will all depend on how well users use it. But I think that in the long term Google will be able to do absolutely nothing to combat the use (good use) of AIs and therefore will not penalize or reduce the ranking of those who generate valuable content.

 

[ianovici]

My opinion is totally different.

AI sites work great, don't own one but recently found out some people who have networks and they are killing it. One of them has one that made 43k last month(ezoic + amazon aff). I guess there are more people who make serious money with it.

The sites are really basic, with no extra categories, just a "Blog" category, where they dump everything possible. The articles are rudimentary formatted, images may be completely unrelated, and the headers and so on. I would do everything differently, but again, they are killing it. No contact, no nothing. Some of these sites have been there for 1-2 years. No penalty.

They were building them with articleforge and something else. Now they switched completely to chatgpt. The content is in all the niches possible, from medical, and credit to f****n horoscope and "best shemale tube". All are on the same category and site. The top earner has been in serps for over a year, with no penalty, no nothing. 200k visitors daily.

I'm starting to reevaluate my strategy and think differently about EEAT. I'm starting to think that EEAT is different on the algorithm level, and not the same as they publicly make it look.

I've looked for all the possible things, tried to make sense of "why do they rank", and nothing. No interlinking, no f***n nothing. Not even "Recent Posts". Bare structure. The only thing I've found, is really, really, nice domains. Super strong. So strong that they can rank relatively fast on medium-difficult keys.

I'm gonna wait until the current update is done, see where I land with my projects, and take it from there. Gonna try to look at things differently and set a new plan in motion.

I have 300 posts waiting to be published, which I've delayed because I'm a little uncertain what is the best course of action. Start new projects or put them in the current network? If I put them in the current network and they don't gain traction, I lose money.

 

[jumper]

Building static pages like "What is a stock" is meant for AI. I would think personal articles on present day events or in-depth tutorials would keep the readers revisiting the site. With all the advanced GPT3,4 and J coming out parsing the artificial and human would be next to impossible. This would be perfect for affiliate sites I would think.

The leverage given to pump out thousands of articles should make money. I don't know what Google will do with the new type of spam to be honest.

 

[TheAbadiKiller]

I've looked for all the possible things, tried to make sense of "why do they rank", and nothing. No interlinking, no f***n nothing. Not even "Recent Posts". Bare structure. The only thing I've found, is really, really, nice domains. Super strong. So strong that they can rank relatively fast on medium-difficult keys.

I've analyzed some of these sites. The common pattern of those websites doing well is a strong backlink profile. Usually, auction domains or repurposed expired domains.

 

[bubududu]

They were building them with articleforge and something else. Now they switched completely to chatgpt. The content is in all the niches possible, from medical, and credit to f****n horoscope and "best shemale tube". All are on the same category and site. The top earner has been in serps for over a year, with no penalty, no nothing. 200k visitors daily.

I know some of these guys. And I can strongly believe reveal to you that they only have one big secret for all of those achivements that you admire. That is a high DA expired domain that they winned on Godaddy auction. Most of them are old restaurant with maps & GMB you can see when search the brand name.

 

[Zoro]

I've analyzed some of these sites. The common pattern of those websites doing well is a strong backlink profile. Usually, auction domains or repurposed expired domains.

Hello Captain!

Are we saying backlinks mather that much?
I thought that was a 2022 kind of thing, look at Adam Enfroy.

 

[Dreamchaser]

Thanks Alex. Is there a specific verification tool you're using? I've tested using ZeroGPT and Open AI's own classifier. I found ZeroGPT the most accurate.

I've found Draft and Goal to be quite accurate so far in detecting ai writing. Better than most of the ai detectors in the market even though it is still in the beta stage.

Give it a spin. Their website is detector.dng.ai.

 

[Thonglor]

I think Google has been very clear about one thing and has tried to insinuate another.

The thing they're clear about is AI content is okay, you won't get penalized. It just needs to be for the reader first and needs to be high quality, as it will be judged in that fashion like any other content.

What they're saying between the lines if you read all the documentation is that there is an expectation that you ADMIT that your content is written by AI.

Here is where my opinion about the future of this comes in:

They'll know it is, and if you don't admit to it, that's going to be like a zero multiplier against your score, or some really low number. You'll have an anchor tied to your legs and won't rank well no matter how good the content is. Why? Because of EEAT and the expectation that there is an author. And if you claim "Johnny Two Shoes" is the author when Google knows it's AI, then that's zero EEAT.

But if you say "this content was produced by AI and fact checked, reviewed, edited, and punched up by Johnny" then you won't get a zero multiplier. Because you've told the users what they need to know, as in WHO wrote the content. My opinion is, if you're going to do AI content, this is the way to do it.

My other opinion is you'll still get a "less-than-one" multiplier on your overall ranking score, meaning you don't have an anchor tied to your legs, but maybe a few bricks as you swim up towards the surface of the SERPs. Why? Because EEAT, which is the solution to all this AI content. An AI can only take old content it was trained on and copy it, it can't bring any new information to the table. I can't think of the technical term but there's no new "information load or expansion or addition". It's the same old crap.

And it's not written by anyone this time. How much credit should a team of doctors get if they just copy and pasted information from other sites and signed off on it saying "okay this is accurate". They should get zero credit and it should go to the original sites. That's the conundrum.

So anyways, do you need to cover your tracks as in hide footprints and not let Google know you own more than one of these sites or whatever? No, they've explicitly said it's okay. That doesn't mean you'll rank, though.

Also you won't hide it from them. They won't find you algorithmically or care, but if they wanted to catch you they would. Their cookies are everywhere, you're probably loading Google Fonts on your sites, using Sheets or Docs, Gmail, Adsense, Double Click for Publishers is being used by everyone ad network, so on and so forth. They do go to these lengths for very vindictive manual penalties (I've had one), but they won't do that for AI sites. But my point is if they want you, they've got you already. お前 和 網 死んでいる.

Thanks for your insight and sorry for the late reply. Was in a country where BUSO is pseudo-banned just like Twitter. Didn't think that would happen!

I'm shocked that G would go all out to penalize you. Until now, I just assumed they wouldn't spend that much resources on a single publisher. Would love to hear more about what happened if you're willing to share. Either way, I hope you recovered.

My gut tells me that Google can't confidently tell what's AI generated (yet) so the only way they can be sure is using some sort of confidence interval or probability calculation. The degree to which your content skews towards 'not helpful AI content' will determine if you get anchors, cinder blocks, bricks or pebbles tied to your legs. EEAT, new ideas or concepts, original images etc. will likely play into the 'helpful' part.

Back to the crux of the topic. I'm not going through the hassle of hiding anything. Like you mentioned - if they know, they know. But I'm going to make sure I'm amping up the 'helpful' bit as much as I can while keeping this project as scalable as possible.

100% agree with Ryuzaki. I agree that if they want to find you, they will find you.

But I reflect and go a bit further, more long term.

The ways in which we users are using AI are still very simple, even the AI services themselves are very new (just a few months ago we didn't have ChatGPT).

I believe that not too long from now AIs will be able to generate valuable content, not only with old content.

If they connect to the Internet and from the information we give them, they will generate new ideas, conclusions, etc...

If we, the users, pass raw (new) data to the AI so that it can generate new ideas, texts, conclusions, the game changes. A simple example: we give the AI hundreds of real opinions of users who have bought a certain product and we ask it to generate an opinion or a review based on this information.

Another example: a journalist gives the AI the highlights of an event or an investigation and asks the AI to help write a news story that is as easy to understand as possible.

The possibilities are endless. It will all depend on how well users use it. But I think that in the long term Google will be able to do absolutely nothing to combat the use (good use) of AIs and therefore will not penalize or reduce the ranking of those who generate valuable content.

Totally agree with your thoughts on the future. Sundar Pichai himself admits that the tech will outpace societal norms so creating some sort of regulation around it will get increasingly harder. Pair that and tech companies laying off employees en masse.

Over time, you would wonder how Google and other companies can keep control of this, much less draw the fine line between what content should be penalized or not. The only exception is if you are obviously generating not helpful and terrible content (a la Conch House).

My opinion is totally different.

AI sites work great, don't own one but recently found out some people who have networks and they are killing it. One of them has one that made 43k last month(ezoic + amazon aff). I guess there are more people who make serious money with it.

The sites are really basic, with no extra categories, just a "Blog" category, where they dump everything possible. The articles are rudimentary formatted, images may be completely unrelated, and the headers and so on. I would do everything differently, but again, they are killing it. No contact, no nothing. Some of these sites have been there for 1-2 years. No penalty.

They were building them with articleforge and something else. Now they switched completely to chatgpt. The content is in all the niches possible, from medical, and credit to f****n horoscope and "best shemale tube". All are on the same category and site. The top earner has been in serps for over a year, with no penalty, no nothing. 200k visitors daily.

I'm starting to reevaluate my strategy and think differently about EEAT. I'm starting to think that EEAT is different on the algorithm level, and not the same as they publicly make it look.

I've looked for all the possible things, tried to make sense of "why do they rank", and nothing. No interlinking, no f***n nothing. Not even "Recent Posts". Bare structure. The only thing I've found, is really, really, nice domains. Super strong. So strong that they can rank relatively fast on medium-difficult keys.

I'm gonna wait until the current update is done, see where I land with my projects, and take it from there. Gonna try to look at things differently and set a new plan in motion.

I have 300 posts waiting to be published, which I've delayed because I'm a little uncertain what is the best course of action. Start new projects or put them in the current network? If I put them in the current network and they don't gain traction, I lose money.

Definitely an interesting perspective on this. I'd be curious to see how long their staying power is because to me it's a bit of a churn and burn model.

But at 45k a month and they been running this for at least a year now? I can see why they do it.

This would be a prime example where I'd try to 'cover my tracks' as much as I can. Your AdSense account could totally get banned for some of the um more 'eclectic' content. That being said, they're are probably buying accounts at that stage.