Ever since ChatGPT and other AI content creation launched into the mainstream, one question has been on the tip of many users' tongues.
Can my audience, AI content detector, or Google detect AI content, and if so, is there a way to get around AI detector algorithms?
The world of AI is still relatively unknown, with many of us starting to scratch the surface of what is capable and what isn't with AI content and other capabilities.
For some, it presents lots of opportunities for legitimate growth and productivity. For others, it represents an opportunity to earn a quick buck and try to con their audience or the Google algorithm.
In recent updates, Google has stated that it has no penalties in place for AI-generated content, stating that it has no issue with how content is produced so long as it is deemed unique and insightful for its users.
And that is a key statement to study. While Google may be fine with letting AI content through its detection tool (it has to be, really, given its recent launch of Bard!), it still wants to see content that is highly unique and offers a fresh perspective on a topic.
However, while some content creators are utilizing the opportunities on offer via AI content creation tools alongside excellent insight and human writing, others are trying to cheat the algorithm by using AI tools that claim to bypass AI content detection tools.
Sure, there can be small instances of false positives, but in general, AI detectors like Originality.ai quickly pick up 100% AI-generated content.
And as per our recent case study, that can have some significant impacts on your results!
In recent times, many people attempted to use Quillbot to bypass AI detection, but now that Originality.ai have successfully created software that identifies AI paraphrasing, they are searching for other tools to avoid AI content detection.
But do they actually work?
Here, we will review four of the most prominent tools that claim to turn AI-generated text into what AI detectors deem as human-written content to see whether they live up to the claims they make.
Price: $0 - $200 per month
Tagline: "Concerned at getting caught by Originality.ai or other content detectors? Use our tool to avoid detection and do better in search engines."
Firstly, let's look at CogniBypass. CogniBypass is a particularly interesting tool, as not only does it claim to paraphrase content making it undetectable ai generated content, but it also has its own version of ChatGPT.
However, unlike ChatGPT, the CogniGPT tool claims to produce content that is undetectable by AI content detectors the first time around, with no paraphrasing required.
But is that the case?
To put this to the test, we decided to test both the CogniBypass and CogniGPT tools, using the same prompt on ChatGPT to see the difference and review whether they work as they promise to their target audience.
For this test, let's keep the content simple, acting as a fake dental practice, talking about the importance of brushing your teeth.
Here's our prompt:
"Please write an article on the importance of brushing your teeth well for my dental practice. Make the content sound authoritative, and add sections surrounding toothpaste, toothbrushes, and frequency of brushing, as well as the issues if you don't brush your teeth regularly. In the conclusion, please promote our own brand toothbrush, the brusher1000."
Here's what ChatGPT gave us and how Originality.ai scored that content:
And here's what CogniGPT gave us and how Originality.ai scored that content:
At first glance, the tool works perfectly, scoring a 97% human-written score on Originality.ai.
But let's look in a little more detail at the content itself...
From just the introduction, it is clear immediately how this content has gotten such a high human score in Originality.ai.
It doesn't make any sense!
Phrases like:
"Yet not all cleaning regimens are developed equivalent."
Cleaning your teeth well not just assists keep a spectacular smile yet it likewise plays a vital duty in stopping numerous dental health and wellness concerns."
"Tooth brushes are the devices that bring the tooth paste and your teeth with each other."
Once the content is tidied up to make grammatical sense, it reverts back to 100% AI-generated content on Originality.ai.
Let's quickly go back to our ChatGPT version. In its current iteration, the content is flagged as 100% AI-generated content, so let's run it through the CogniBypass tool to see what impact that has.
Disclaimer: Annoyingly, the tool only allows us to do 150 words at a time!
Now that same content is showing as 51% human-written, 49% ai-written, but let's check the actual content again.
Again, there are clear issues with the content created, with lines such as:
"Excellent dental health is critical in preserving a healthy and balanced and also glowing smile."
It is clear that CogniBypass focuses on broken English and poor grammar to trick AI detection tools, which, if you are a native English speaker, is easy to identify.
Price: $0 - $600 per month
Tagline: "Try the best AI blog writing tool on the market, and generate complete, non-detectable articles with outlines, blog content, keywords, and Call To Action in a minute – the first blog is on the house!"
The next tool to look at that claims to bypass AI detectors is blogassistant. Let's see if we have similar results as CogniBypass.
While it is the most expensive option on this list of AI detection bypassing tools, it does offer the first blog for free, so let's take advantage of that and test it out!
Here's the prompt we gave blogassistant, to see if it can avoid AI detection:
"Please write me an article on climate change and why new houses should all be fitted with solar panels, free of charge."
blogassistant has a much more detailed system in place, asking us to fill out much more information before getting started.
Now let's put this into Originality.ai:
Once again, this seems like a good result, but based on the previous results, let's check the grammar on the information before we get ahead of ourselves.
As expected, the reason this content avoids AI detection is due to several spelling mistakes and grammatical errors. In fact, in the first section alone, Grammarly detects unnecessary phrases, missing commas, and missing words ten times.
The reviews for this tool also reflect this sentiment.
in one review from LearnWire, the reviewer states the following:
"I did the trial and realised that their trick to be undetectable by AI trackers is to add a lot of grammar mistakes and blatant punctuation errors. If you fix these errors, Originality AI marks the text as 99 % AI. So, they generate bad-quality text on purpose to bypass AI detectors, but the text is easily detected as AI once you fix these errors. It is quite silly. Too bad that AppSumo allows such a bad product on its site. I guess they have good reviews because they will give free words or some incentive."
Price: $12 - $349 per month
Tagline: "A tool that removes AI content by paraphrasing"
The next tool to look at that claims to bypass AI detection tools is GPTinf. With GPTing, you can register for free to use a limited version of the tool, and then the monthly cost depends on how many words you use per month.
The tool offers to rephrase your AI text so that it beats AI detectors like Originality.ai, and the platform is very simplistic and easy to use.
But does it work?
As with the other tools on this list, let's test it out. This time, let's pretend we run a coffee review website in downtown Manhattan, and we want to create an article on the best places to get coffee from in the local area. We also want to promote our article on email to our subscribers and promote our new coffee at the same time.
Let's start with the blog article first again. Here's our prompt for ChatGPT:
"Please write me an article reviewing the five best coffee shops in downtown Manhattan. The article must come from a place of authority, as we are a coffee review website called Cool Beans. Please review the taste, price, and location while also incorporating customer reviews for each of the five."
Here's what we got in response:
Now let's quickly run that through Originality.ai:
Yep, all good there! Now let's paraphrase it in GPTinf:
Now, let's run our paraphrased content through Originality.ai:
As you can see, the content is still flagged as 100% AI. However, it is easy to see here how GPTinf could work, but not because the content is in a human-written style. There are actually a ton of spelling mistakes, strange phrases, and a lack of spaces between words. It is clear that GPTinf uses this method to try and trick tools, but once you tidy up the grammar, it immediately flags as AI-written.
Now let's look at the product description. Here's our prompt, continuing on ChatGPT:
"Use this writing style and information to create a product description for Cool Beans launching their very first unique coffee beans that are ethically sourced, high-quality, and available to pre-order for 30% off, using code ILOVEBEANS"
Here are the results:
And here's what Originality.ai says:
Here's what GPTinf paraphrase it to:
And here's what Originality.ai says to that:
Nice try, GPTinf!
Okay, now let's check out a marketing email. We have our brand new blog we want to promote, and we have our coffee beans to sell, so let's create an email in ChatGPT and see if GPTinf can tweak it to get past Originality.ai.
Here's our prompt:
"Turn all of that information and the blog article previously into a sales email. Start by promoting the brand new blog and the sub-promotion of our new beans, available for pre-order"
Here's the result:
And here's what Originality.ai thinks:
No surprises there! Now let's see what GPTinf provide us with:
What does Originality.ai think to those changes?
As with the article and the product description, Originality.ai quickly detects this is still 100% AI-generated content.
Alongside our own research, let's look at what the reviews for GPTinf say.
Price: $0 - $199 per month
Tagline: "Provides anonymity and helps avoid detection on websites."
Finally, we move on to Undetectable, the tool that claims to provide users with anonymity online and avoid detection.
Undetectable.ai is another tool that claims to bypass AI detectors such as Originality.ai, and offers you a simple breakdown of how it will do it for you.
Firstly, you are asked to input your AI-generated content, which could be from any AI writing software, such as ChatGPT or a paid subscription. That content could be a blog post, eBook chapter, product description, or even an email.
Next, the Undetectable.ai tool suggests it uses "cutting-edge" machine learning algorithms to analyze the content you submitted, evaluate the context and intent, and then rewrite it so that it becomes undetectable to AI tools like Originality.ai.
The tool also claims to produce human-like content that search engines will love and that is indistinguishable from actual human writing.
Pretty big claims!
So we decided to put it to the test, with ChatGPT as our AI tool of choice. Using an example company, we decided to create one fully ai-generated blog post, one product description, and one email.
The company we decided to create was a standard eCommerce brand selling 5-panel caps. Our made-up company wanted to write an article on why caps are great for when you're travelling and promote a new product across email.
Here is the prompt we used for this test:
"Write me an article on the benefits of wearing a 5-panel cap while travelling. Start with a short introduction to the topic, then the benefits, and then a conclusion. Within the benefits, reference three made-up caps, called the traveller, the recluse, and the hermit. Highlight the benefits of each made-up product seamlessly within the article. In the conclusion, reference a new 5-panel cap that is landing soon, called the pioneer, and keep your eyes peeled for more details. Keep the tone light and adventurous, and be sure to mention our brand "Kappies" throughout"
And here's what ChatGPT provided us with:
Those are pretty strong foundations for a great article! Now let's check if Originality.ai can identify this as AI-generated.
As expected, Originality.ai immediately detects that it is 100% AI-generated content, as many other AI content detection tools would as well.
So let's put Undetectable.ai to the test, seeing if it can trick AI content detectors and make it seem like human-written text.
As you can see, Originality.ai still shows the results are 100% AI-generated, disproving the claims made by Undetectable.ai.
Next, let's focus on the product description. As mentioned in our blog prompt, we are launching a new 5-panel cap, so let's get a product description created on ChatGPT.
Here's our prompt:
"Write a product description for The Pioneer 5-panel cap. Made for travellers, made with sustainable resources, limited-time free shipping, and high-quality and stylish"
And here are the results:
Now let's check that in Originality.ai for AI detection:
As hoped, Originality.ai instantly registers this as AI-generated content. Now let's check it after running it through Undetectable.ai.
Once again, Originality.ai immediately registers this as 100% AI-produced content creation, further proof of the strength of this AI detection tool.
But what about when we use Undetectable.ai for a promotional email regarding our new product? Let's work one up in ChatGPT and find out.
Here's the prompt we used:
"Turn this information into a promotional email, launching the new product with the limited-time free shipping offer, and use code "Kappies" for 10% off. offer lasts three days"
And here's what ChatGPT offered us:
Now let's quickly check that in Originality.ai to see if it identifies it as AI-generated output:
As expected. Now let's run it through Undetectable.ai to see if it can create content that Originality.ai thinks human writers produced.
Once again, there's no tricking Originality.ai that this is human-written content! Even with a different sentence structure, Ai writing assistant tools struggle to get AI-generated content past Originality.ai.
Given our findings, we were intrigued to see some of the reviews from users to see how well the tool was received by users.
However, there are also some positive reviews, so it is clear that the tool works for some users:
Price: $99 - $299 per month
Tagline: "From emails to blogs to papers – Our Undetectable AI, plagiarism-free AI tool allows you to get more done, 10x faster so you can concentrate on what matters."
Another tool that has gained traction in recent months is StealthGPT, and it’s a tool that we have reviewed in greater detail as well.
In it’s marketing, StealthGPT suggests it can break down the anti-AI stigma, by generating content that will bypass any and all AI content detectors.
Well, I guess that includes Originality.ai, so let’s see if those claims are true by putting to the test.
To do this, let’s create an imaginary clothing brand called Did I Mention Clothing. The brand will sell clothing that promotes championing your achievements and the benefits of a little humble brag from time to time.
Let’s imagine the company is bringing out a new collection called the “Did I Mention I Ran a Marathon?”. To promote the product, we need to create a blog article, product description, and email.
Let’s see if StealthGPT can create high-quality, AI-bypassing content.
Let’s start by creating our blog article, and launching the new product. We will start by creating it in ChatGPT, running it through Originality.ai for a benchmark detection score, and then doing the same after running it through StealthGPT.
Here’s our prompt:
“Compose a captivating and informative blog article to introduce the latest collection from Did I Mention Clothing, titled "Did I Mention I Ran a Marathon?". The article should highlight the key features of the new clothing line, emphasizing the brand's ethos of celebrating personal achievements and the art of the humble brag.”
Here’s our ChatGPT content:
And here’s what Originality.ai says:
Now what does it say after it goes through StealthGPT:
Okay so no luck there, but what about with a product description in the same ilk?
Here’s our ChatGPT prompt:
“Create engaging and compelling product descriptions for the key items in the "Did I Mention I Ran a Marathon?" collection from Did I Mention Clothing. The collection includes a hoodie, tee, beanie, and shorts. Emphasize the unique features of each item, highlighting their design, functionality, and how they embody the brand's ethos of celebrating personal achievements and the art of the humble brag. Craft the descriptions with a mix of enthusiasm, style, and a touch of the brand's distinct personality. Make sure to inspire the reader to envision themselves confidently showcasing their marathon triumphs in these exceptional pieces.”
Here’s what we got:
Now, how does that look in Originality.ai?
As expected, Originality.ai instantly detects this content as AI-written. But what about after we run in through StealthGPT?
As you can see, now the content is 100% human generated based on the results in Originality.ai, which is perfect, right?
Well, not quite.
As with other AI bypassing tools, StealthGPT has managed to achieve this score by making the content almost incoherent. For example, here’s that same copy in Grammarly:
As you can see, the content is riddled with grammatical errors. Here are some of the most erroneous ones:
So while the content passes our detector, it certainly won’t resonate with your audience.
Finally, let’s look at an email example, promoting the same clothing collection.
Here’s our prompt:
“Compose an engaging and persuasive email to promote the "Did I Mention I Ran a Marathon?" collection from Did I Mention Clothing. The email should highlight the key items in the collection (hoodie, tee, beanie, and shorts), emphasizing their unique features, design, and how they embody the brand's philosophy of celebrating personal achievements and the art of the humble brag. Craft the email with a friendly and enthusiastic tone, encouraging readers to explore the collection, envision themselves proudly showcasing their marathon triumphs, and make a purchase. Conclude with a compelling call to action to visit the website.”
And here’s what ChatGPT gave us:
And here’s what Originality.ai said about that:
And here’s the score after running it through StealthGPT:
However once again we come up against the same issues, with the tool making the content almost incoherent try and pass detectors. Here are some of the worst examples from the text:
Finally, let’s take a look at some real customer reviews from Trustpilot:
Price: $200 - $500 per month
Tagline: "Stealthwriter rewrites AI content into human-like text. Guaranteed plagiarism-free & undetectable by AI detectors."
Next, let’s look at Stealthwriter, another AI bypassing tool, not to be confused with the similarly named, StealthGPT.
Stealhwriter makes all the same claims as the other AI bypassing tools on this list, and it’s a tool we have reviewed before. So let’s put it through the same tests and see how it gets on.
This time, we will create an imaginary computer company that sells a range of PCs, laptops, and computer accessories, called “PC Parts.”
Let’s start with our blog article test. Here is our prompt:
“Generate a captivating and informative blog article for the imaginary computer company, "PC Parts." The article should highlight the diverse range of PCs, laptops, and computer accessories offered by the brand. Emphasize the key features, cutting-edge technology, and the commitment to providing top-notch performance and user experience. Additionally, touch upon the brand's mission, customer-centric approach, and any upcoming innovations in the world of PCs and computer technology. Craft the article to resonate with tech enthusiasts, casual users, and anyone seeking reliable and innovative computing solutions.”
Here’s the result:
And here’s what Originality.ai thinks:
And here’s how it performs after we run it through StealthWriter:
But as we now know, simply looking at this percentage score is not enough to determine whether the tool has worked or not, and as with other AI bypass tools, the actual content is basically gibberish:
With some of the most glaring examples being:
Now let’s do the same for a product description.
“Compose an engaging and informative product description for the latest PC keyboard from PC Parts. The keyboard is designed to offer a seamless and elevated typing experience, catering to both gaming enthusiasts and productivity-focused users. Emphasize the key features, including any innovative design elements, customizable options, and ergonomic considerations. Capture the essence of the product, highlighting how it aligns with PC Parts' commitment to providing cutting-edge technology and enhancing the user's computing journey. Craft the description with a blend of enthusiasm, technical detail, and a touch of the brand's commitment to user satisfaction.”
Here’s what Originality.ai says to that:
And here’s what Originality.ai says after running the copy through StealthWriter:
Again, let’s check some of the grammar issues that lead to this score:
Finally, let’s create an email to launch our new keyboard, and see what Stealthwriter can offer us (although I think we know the answer by now!)
Here’s our prompt:
“Compose a compelling and persuasive email to introduce the PC Parts QuantumFlow Pro Keyboard. The email should highlight the key features of the keyboard, emphasizing its suitability for both gaming enthusiasts and productivity-focused users. Craft the email with a tone of excitement and enthusiasm, showcasing the innovative design elements, customizable options, and ergonomic considerations that make the QuantumFlow Pro Keyboard stand out. Conclude with a strong call to action, encouraging readers to explore the product on the PC Parts website and make a purchase. Infuse the email with the brand's commitment to cutting-edge technology and user satisfaction.”
Here’s what Originality.ai thinks of the original ChatGPT content:
And here’s the score after running it through Stealth Writer:
…and here’s why it gets that score:
Finally, let’s take a look at some real customer reviews from Trust Pilot:
Price: $0 - $1500 per month
Tagline: "Save time on your assignments, boost grades, and easily bypass AI detectors like TurnItIn and GPTZero."
The next bypassing tool we will look at is Phrasly AI. Like the other tools on this list, it claims to create content that bypasses tools like ours while also providing you with excellent content. We have reviewed the tool before, so let’s run it through the same tests as the others on this list, using the same PC Parts business for our test.
As we already know, our ChatGPT blog version flags up as 100% AI-generated, but what about if we run it through Phrasly AI?
However, we have the same issue here as with other “bypass” tools, in that the new content doesn’t actually make sense:
I wouldn’t want to speak for everyone, but the term “womb of innovation” just makes me feel a bit uneasy!
Now let’s do the same with our PC Parts product description, although I imagine we will see similar results.
Similar score, but similar issues:
At least there was no reference to the womb of innovation this time!
Finally, let’s test out an email, although at this point we all know what the outcome will be.
The content scores as expected, and for the reasons we expected as well:
Price: $0 - $188 per month
Tagline: "Bypass AI detectors, and transform AI-generated content to humanlike copy with our free tool."
Finally, let’s take a look at another AI tool that claims to bypass AI detectors. It is also another tool that we have reviewed in great detail, as it claims to create content that passes our detectors.
But let’s see if that is the case or not, by setting up another test, similar to the ones before. This time, we will pretend to own a skateboarding shop, that sells everything you could need, from skateboards, accessories, safety gear, and clothing.
Firstly, let’s start by creating a blog post for our skateboarding brand in ChatGPT.
Here’s our prompt:
“Compose an exciting and informative blog post for our skateboarding shop, highlighting the latest additions to our skateboard collection. Emphasize the key features, unique designs, and innovative technologies of the new skateboards. Capture the essence of the skateboarding culture, catering to both beginners and seasoned riders. Incorporate details about the materials used, performance benefits, and any collaborations or inspirations behind the skateboard designs. Infuse the post with enthusiasm, encouraging readers to explore the thrilling world of skateboarding with our cutting-edge boards. Don't forget to mention safety gear, accessories, and clothing that complement the skateboards. Craft the article with a tone that resonates with the passion and excitement of the skateboarding community.”
Here’s the Originality.ai score:
And here’s the score after adapting the content in Hidemy AI:
Even with the changes, Originality.ai detects the content as AI written, despite the quality of the content dropping.
Now, let’s do the same with our product description test. Here’s the prompt for ChatGPT:
“Create an engaging and dynamic product description for our latest skateboard collection at [Your Skateboarding Shop]. Highlight the key features, unique designs, and innovation incorporated into these boards. Infuse the description with the vibrant spirit of skateboarding culture, emphasizing how each board caters to the diverse styles of riders. Include details about the materials used, artistic inspirations, and any collaborative efforts that make these skateboards stand out. Craft the description with a blend of enthusiasm, technical detail, and a genuine passion for the skateboarding community. Conclude with a strong call to action, encouraging readers to explore the collection on the [Your Skateboarding Shop] website and find their next ride.”
Here’s what Originality.ai thinks about that content:
And now let’s see that score after running the content through Hidemy AI:
As with the blog example, Originality.ai still recognizes the content as AI-written, despite the changes.
And finally, let’s test out an email in the same fashion.
Here’s the Originality.ai score without any changes:
And here it is after running it through Hidemy AI:
When it comes to creating content that bypasses AI detectors, there are two clear ways to do so.
AI content generation tools certainly offer many benefits and help you generate content at scale. However, if you want your audience to read your content and become engaged with its uniqueness and authenticity, then you still need to add the touch and flair of human creativity.
Although this is just a small sample size, a more complete study would be needed to truly understand the tool's capabilities at bypassing AI detection. However, from this initial research, it is clear that these tools do not currently outsmart the capabilities of Originality.ai.
Achieving writing that is entirely undetectable by AI is not possible without resorting to bad grammar or making significant changes that may render the text nonsensical to humans.
There is no AI writer that can guarantee complete undetectability as advanced language models and detection algorithms continue to evolve.
Yes, our studies have shown that Google can detect AI spam and despite up to 50% of the web already being AI content there is only a small amount of AI content in Googles results. Using undetectable tools which attempt to make the content undetectable by adding errors and weird formatting is not recommended. There are other advanced SEO strategies for AI content.
Bypassing AI text detection without utilizing bad grammar or incorporating human writing is not currently feasible, as sophisticated algorithms are designed to identify patterns indicative of AI-generated content.
Yes, AI-written content can be detected through various methods, such as using language model classifiers or other AI-based detection algorithms designed to distinguish between human-written and AI-generated text.
