✅ How to create 100% unique and undetectable articles using Chat-GPT.

nirajjoshi

nirajjoshi

Power Member
Joined
Sep 20, 2015
Messages
576
Reaction score
970
I will show you how to create human like, undetectable and unique articles using Chat-GPT.

If you don't know yet, there are tools to detect Chat-GPT content and they are really accurate.

Here are 2 tools that are really good to detect Chat-GPT text.
https://contentatscale.ai/ai-content-detector/
https://www.zerogpt.com/


What is this method?
Chat-GPT generated text that is too professional. We have to keep everyone in mind when we post articles online. It's too hard for a non-native English speaker or kids to understand chat-gpt generated articles. Because it spits out a lot of high grade English words. It's uncommon for a human writer to use such an advanced level of English unless they are publishing some kind of research-paper or scientific journal.

I will show you the example of detection.
My generated text using Chat-GPT

I have colored the texts that are the problem with ai written text.
Prompt:
Write a youtube description for the video title "How To Make Money Online"

Are you looking for ways to make money online but don't know where to start? This video is for you! In this comprehensive guide, we will cover various methods to make money online, including freelance work, online surveys, affiliate marketing, and more.

We'll provide tips and tricks for each method, as well as share our personal experiences with making money online. Whether you're a stay-at-home parent, a student, or simply looking to earn some extra income, this video is packed with valuable information to help you get started.

Don't waste any more time searching for ways to make money online. Watch this video now and start your journey to financial freedom!
Click to expand...

These few words ruin the entire text. Just changing those words don't work.

Results:
Contentscale
Content Scale.jpeg

ZeroGPT
ZEROGPT.jpeg

You might want to consider checking your articles if you directly copy them from Chat-GPT. You might be hurting your clients.

The Method

Step 1: Generate the text using any prompt. I have tried every single prompt. From this forum and the AIRPM extention too. Every generated text is detected.

My generated text is given above. It is detected as AI written.

Step 2: Copy it to hemmingway app.

hemmingway.jpeg
Check out the readability errors. We have to fix them using our own words! But worry not, we will take help from chat-Gpt.

Step 3: Using the following prompt to re-write the text. We will use it as a ref to re-write the article in our own words.
Prompt:
Expand the following text a little bit and write it so a 7 year old kid can understand it. "Text"
Click to expand...

Generated Text
Hey there kiddo! Do you know that people can make money by working on their computers and the internet? Well, in this video, we are going to show you different ways to do it!

For example, some people can work for others from home and earn money doing things like writing, designing or programming. Others can answer online surveys and get paid for their opinions! There's even a way called "affiliate marketing" where people can recommend things they like and get a commission if someone buys it.

We will give you tips and tricks for each method, and tell you stories about how we made money online. You don't need to be a grown-up to make money online, even if you're a kid, you can make some extra money too!

So, don't worry if you don't know where to start, we got you covered! Just watch this video and you'll learn how to make money online. It's a great way to earn some extra cash and have more financial freedom!
Click to expand...

Step 4: Paste this text into the hemmingway app and start re-writing the article using your own words.

kidified.jpeg

Step 5: Write the article in simple words. Take help from the expanded version we just created. Here's my text.
Do you know that it's possible to earn money online using your computer and internet? In this video we are going to show you different techniques to earn money online.

Article writing, graphics desigining or programming are some of the examples. There are also online surveys and tasks you can do to earn money online.

Advanced techniques like affiliate marketing are also explained in this video.

In this video you will learn various tips and tricks about how we make thousands of dollars online. Even if you're a beginner, you will be able to make some money after watching this video.

Watch the entire video to learn online money making techniques. Pave a way for your finiancial freedom.
Click to expand...
MYVERSION.jpeg

Step 6: Now paste this into the AI detector. Now the score of detection will be less. It's time to re-purify the article. You just need to edit text that are highlighted.
detection2.jpeg

detection3.jpeg

Step 7: Open Quillbot and paraphrase the highlighted text. Use fluency option and minimal settings. Re-write hard to read sentences too! Like the below example.
Hard to read
Even if you're a beginner, after watching this video, you'll be able to make some money.
Click to expand...
Easy to read
You'll be able to earn money after watching this video.
Click to expand...

human1.jpeg
human2.jpeg
 
Advertise on BHW
Great post, but these AI detectors are not as accurate as you might think.

Here is a study, which finds "In this paper, both empirically and theoretically, we show that these detectors are not reliable in practical scenarios. Empirically, we show that paraphrasing attacks, where a light paraphraser is applied on top of the generative text model, can break a whole range of detectors, including the ones using the watermarking schemes as well as neural network-based detectors and zero-shot classifiers. We then provide a theoretical impossibility result indicating that for a sufficiently good language model, even the best-possible detector can only perform marginally better than a random classifier. Finally, we show that even LLMs protected by watermarking schemes can be vulnerable against spoofing attacks where adversarial humans can infer hidden watermarking signatures and add them to their generated text to be detected as text generated by the LLMs, potentially causing reputational damages to their developers. We believe these results can open an honest conversation in the community regarding the ethical and reliable use of AI-generated text."

Also, this screenshot shows the Bible showing up as AI generated, take that as you will.
342902517_756294869458571_1588715008068490304_n.jpg


In the end, you might want to get a lower detection score simply because clients demand it but if you're uploading content for you own websites, I wouldnt care less. If the content is helpful, valuable and satisfies searcher intent then AI detection score does not matter.
 
Laaaawan said:
That's cool
Click to expand...

Nikolaos said:
Good method!
Click to expand...

Thanks a lot!

StripeGee said:
What of if you use Chatgpt api
Click to expand...

I don't think it can be done on automated scale. After all we're giving it a human touch.

FromAfrica said:
Nice share

Thanks
Click to expand...

Thanks!

CampData said:
Would love to see the same method for other Latin languages (Spanish, French, Italian...)
Click to expand...

It can be done but Chat-GPT is terrible in foreign language text generation.

tattooedbuddha said:
Great post, but these AI detectors are not as accurate as you might think.

Here is a study, which finds "In this paper, both empirically and theoretically, we show that these detectors are not reliable in practical scenarios. Empirically, we show that paraphrasing attacks, where a light paraphraser is applied on top of the generative text model, can break a whole range of detectors, including the ones using the watermarking schemes as well as neural network-based detectors and zero-shot classifiers. We then provide a theoretical impossibility result indicating that for a sufficiently good language model, even the best-possible detector can only perform marginally better than a random classifier. Finally, we show that even LLMs protected by watermarking schemes can be vulnerable against spoofing attacks where adversarial humans can infer hidden watermarking signatures and add them to their generated text to be detected as text generated by the LLMs, potentially causing reputational damages to their developers. We believe these results can open an honest conversation in the community regarding the ethical and reliable use of AI-generated text."

Also, this screenshot shows the Bible showing up as AI generated, take that as you will.
342902517_756294869458571_1588715008068490304_n.jpg


In the end, you might want to get a lower detection score simply because clients demand it but if you're uploading content for you own websites, I wouldnt care less. If the content is helpful, valuable and satisfies searcher intent then AI detection score does not matter.
Click to expand...

Interesting take. I agree that the detection technique is flawed but it's currently used by clients to check content.

Currently ChatGPT is only better for generating ideas and using those ideas to create content.

Google partners have manual raters who check content, the reputation of author, the reputation of website and then rate according.

If your website have medical related content then it's very important that the post author is established as a medical professional. That's why now a days medical related articles have "peer reviewed by medical professionals" or similar label.
 
Detectors just suck in general. Haven’t seen a single one give real accurate info. Either they are tweaked so heavy that everything is AI (including fully human written) or they don’t detect any AI beyond the most basic prompts.

Oh and GPT4 in the playground easily passes content at scale out of the box with no tweaks needed and you don’t even have to put “in the style of” or other tweaks.
 
It is clear that not everything is perfect and, of course, there are flaws.
 
nirajjoshi said:
I will show you how to create human like, undetectable and unique articles using Chat-GPT.


An interesting idea, but its implementation requires too many steps and time. You could have written those few paragraphs yourself in 2 minutes.
Click to expand...
 
gplmedialinks said:
Here are a few things you can do to ensure that the articles generated by Chat-GPT are 100% unique and undetectable.

Choose a specific topic: Start by providing a clear and specific topic for the article you want to generate. This will help Chat-GPT to better understand the context and generate more relevant content.

Use original prompts: Instead of copying and pasting existing articles or prompts, try to come up with original prompts and ideas that are not already available online. This will help ensure that the articles generated by Chat-GPT are unique and not plagiarized.

Modify the output: Once you have generated an article using Chat-GPT, review it carefully and make any necessary modifications to ensure that the content is unique and meets your requirements. You can add your own ideas, research, or insights to the article to make it more unique and personalized.

Use plagiarism checker tools: Before publishing the article, it's important to use plagiarism checker tools to ensure that the content is 100% unique and undetectable. These tools can help identify any similarities between the generated content and existing online content, and allow you to make any necessary modifications to ensure the article is truly unique.

Remember, while Chat-GPT can be a powerful tool for generating content, it's important to use it responsibly and ensure that the content you create is high-quality, unique, and ethical.
Click to expand...

Was this generated using ChatGPT? No offence, just asking out of curiosity.
 
  1. Rewrite the following text so that it's harder to detect as coming from ChatGPT by a ChatGPT detector.
  2. Rewrite the following text. Write like a human, more spoken language, more line breaks.
  3. Paraphrase the following, use random words and rare words. Write more like a human.
 
I published many articles some of which are ranking top 3 with chatGPT no content or words editing. How many sites today are writing about same topics there will always be some duplicated words and lines when you check for plagiarism with any tool. I think its not much to worry about it. But if you have time and can fix every article then sure it might be more bulletproof in long term.
 
Why dont you just ask chatgpt to write like a noob without using too complicated words
 
Interesting. I've read something similar in the past
Nice share
 
I consistently get 70-80% human content with my scripts without paraphrase. The Quillbot method works, but imo that sort of reduces the overall output for the human eye. Am still trying to find a way to get 90% all the time without paraphrase.

The trick is to twist the code, to break the patterns.
 
Why even go through all that hassle after Google said AI content is fine if it is useful? Also these "detection" tools even detect the bible or some childrens books as AI content although they have been written long before AI so my personal opinion is these tools are just a clever way to make some additional money from you.
 
Advertise on BHW
You must log in or register to reply here.
Sign up now!

Main Menu

Home Main Forum List Black Hat SEO White Hat SEO BHW Newbie Guide Blogging Black Hat Tools Social Networking Downloads

Marketplace

Content / Copywriting Hosting Images, Logos & Videos Proxies For Sale SEO - Link building SEO - Packages Social Media Social Media - Panels Web Design Misc

Making Money

Affiliate Programs Hire a Freelancer Making Money Pay Per Click Site Flipping

BlackHatWorld

BHW Merch Forum Suggestions & Feedback Introductions Lounge Dispute Resolution

Other

Domain Name Forum IM Journeys Web Hosting Newsletter
Back
Top