We're back with another in-depth dive, this time focusing on StealthGPT.ai, a tool that claims to be the only AI writing platform that can rephrase your content to bypass content detectors.
Whether you're writing essays, blog posts, or detailed marketing PDFs, StealthGPT.ai claims to use unique technology to create high quality AI-generated content that cannot get detected.
StealthGPT was effective at having ChatGPT text get past all AI detectors except Originality.ai's AI Detector. However, once the grammar mistakes which ChatGPT doesn't naturally make were fixed it was easily identified as AI again.
A better writing aid that produces more grammatically correct and undetectable AI content (except does not bypass Originality.ai) is Duey.ai.
Here, we will discuss the following:
StealthGPT claims to take down the anti-AI stigma by producing AI-generated content that bypasses any and all AI content detectors.
The tools claims to help people with college research papers, marketers with long-form content, and just about anyone who wants to generate text-based content.
People are interested in tools like StealthGPT because they are searching for some form of shortcut.
Whether it is due to tight deadlines, a lack of confidence in their work, or an opportunity to save time and effort, many people are interested in the capabilities of AI tools, especially if they can get away with using them without getting caught.
But does StealthGPT.ai actually work? As part of our research, we subscribed for a StealthGPT membership, and decided to put it to the test.
Rather than simply reading the reviews and taking them as gospel, we wanted to dive a little deeper, and put the platform to the test.
For our test, we decided to write a PDF for a made up influencer marketing agency. The idea of the PDF is to provide brands an insight into the benefits of working with social media influencers, in the hope they will then contact the agency to get exclusive access to their list of influencers.
The content is 2,000 words long, and 100% generated within ChatGPT, using detailed AI prompts to create an insightful PDF that adds value.
With our test prepared, we will now run the original version through Originality.ai, GPTZero, Writer, ContentAtScale, and CopyLeaks, to see how it is perceived
We will also run the content through StealthGPT, and then see how that version performs in each tool.
Original content:
StealthGPT version:
Original content:
StealthGPT version:
Original content:
StealthGPT version:
Original content:
StealthGPT version:
Original content:
StealthGPT version
At first glance, the results are not great. For starters, Originality.ai is the only tool to correctly identify that the original content was 100% generated in Chat GPT. That said, ContentAtScale and CopyLeaks also marked it with a high percentage.
On top of that, none of the tools were able to identify the output as 100% AI generated after StealthGPT amended it, with CopyLeaks coming the closest with a 63% AI score.
So how did this occur? After all, StealthGPT does not use human writers to rewrite your content, so why did each tool struggle to identify it as AI-generated?
We decided to look a little deeper, taking the StealthGPT content out of the platform and placing it into Grammarly.
Almost immediately, the results started to make more sense.
For starters, the content score is sitting at 80 out of 100, well below that of a well-written article. A quick glance at the sidebar shows that the content is rather unclear and hard to follow.
A lack of correct grammar and unnecessarily long-winded sentences are both examples of AI detection avoiders trying to trick AI content detectors, but at what price?
The content in its current form is almost unreadable, with no headings or paragraphs in sight. It certainly isn't in a position to upload it to a website, turn it into a PDF, or promote it to clients.
It would take a lot of work to make it structurally correct and would likely result in the AI score shooting straight back up. But that's only an assumption.
So, we did just that.
And how did our newly formatted StealthGPT content fare with the content detectors?
As you can see, the results (apart from Writer) are much more in keeping with what you would expect to see from AI content detectors such as Originality.ai.
You can try out StealthGPT.ai for free, but you can only use 1,000 characters before you are then required to purchase a membership.
StealthGPT.ai attempts to use AI technology and deep learning algorithms to tweak AI-written content so that it is not detected by AI content detections like Originality.ai.
To make undetectable content, you must do more than simply run it through a paraphrasing tool. By combining the results of an AI content platform with well-written human input, you can generate high-quality content, unique blog posts, and articles that will not show up as 100% AI-generated.
However, if you choose to generate 100% of your content with an AI tool, then expect that to be the result when it gets run through an AI content detector like Originality.ai.
At Originality.ai, we believe that AI writing tools certainly have their place in certain businesses, but transparency when using them is vitally important in an era of misinformation.
We certainly don't see ourselves as the anti-AI gatekeepers, but we do believe in the importance of clarity and transparency between clients and businesses, employees and managers, and businesses and their audiences.
A high-quality AI writing platform can produce content with human-like language, and it's important not to mislead or lie to your audience about its source.