HideMyAI vs. Originality.AI: Can You Bypass AI Detection?

The release of ChatGPT started a chain reaction in the AI world. Suddenly, there’s a wide-scale demand for AI tools for specific problems. Whether it’s to finish schoolwork or to help facilitate learning for children with disabilities, AI has been everywhere for the last year.

Of course, as Isaac Newton once said, “every action has an equal and opposite reaction.” As AI tools become more popular, there’s also been an increase in demand for software to govern AI. For example, AI detectors aimed to flag AI writing. One such tool is Originality AI.

However, there’s a new sheriff in town, and it’s not messing around. As more and more people flock to detection tools, it’s becoming apparent that they’re not as accurate as they seem. Therefore, people started looking for ways to bypass AI tools, such as HideMyAI.

But what happens when you lock an AI detector and an AI bypass tool in a room? Which will come out on top?

That’s what I sought to find out in this article.

What Are HideMyAI and Originality.AI?

Originality.AI is a tool that claims to be the most accurate AI content detector in the market. According to their blog, this AI detection tool appears to have a 95.62% accuracy. However, looking through their dataset, there’s one specific type of content they didn’t use to test the accuracy of their application…

And that’s AI bypass tools such as HideMyAI. On the other hand, this tool claims that it can humanize your content so it won’t get flagged by detectors. All you need to do is input your AI-generated content into their editor, pick the right parameters, and wait for the output. It’s all a matter of seconds before you can get undetectable text.

So, what would happen if these two were put to the test against each other? Let’s find out.

HideMyAI vs. Originality.AI: Different Inputs

I’ll only use one AI tool to generate text through the test: ChatGPT. Not only is it free, but it’s also the most accessible and used prompt-based AI generator in the market today.

I’ve divided the tests into different categories, starting with different inputs. I’ll ask ChatGPT to write a formal and casual essay, input the generated text into HideMyAI using the “General Writing'' and “Match To Content” parameters to ensure fairness and paste HideMyAI’s output into Originality’s AI content detector.

Sounds easy, right? So, without further ado…

Test #1: Formal

To start, I asked ChatGPT to write a short, formal essay about Genghis Khan. This is meant to emulate essays that students submit using ChatGPT.

These are the results when I asked HideMyAI to transform the text and what Originality thinks of the new text:

DETECTION AVOIDED! — Likelihood of Being Human-Written: 97%

Test #2: Casual

One of the biggest ways to hide AI content is to inject creativity and casualness into the text. That’s why I also wondered if Originality.AI can detect AI content from a HideMyAI text that was casual to begin with. I asked ChatGPT to write a creative essay about Futurama to test that out.

After transforming the initial text using HideMyAI, this is the score that Originality.AI produced:

DETECTION AVOIDED! — Likelihood of Being Human-Written: 100%

HideMyAI vs. Originality.AI: Different Types

One of the parameters you can adjust in HideMyAI is the text’s use case. This allows the tool to transform your output into something specific, which, in turn, increases its readability. The different use case types in HideMyAI are the following:

• General Writing
• Essay
• News Article
• Blog Post
• Cover Letter
• Marketing Copy
• Email

I’ll test all these except for general writing, as we’ve covered that in the previous section. Moreover, there will only be one prompt for essays, news articles, and blog posts but different ones for cover letters, copies, and emails. This ensures that HideMyAI won’t try to fit a square peg into a round hole by creating emails out of essays, for example.

The prompt for the first part of this section deals with data science and its implications in real-world dilemmas.

This is how well Originality.AI does against HideMyAI for specific use cases:

Test #3: Essay

DETECTION AVOIDED! — Likelihood of Being Human-Written: 73%

Test #4: News Article

DETECTION AVOIDED! — Likelihood of Being Human-Written: 82%

Test #5: Blog Post

DETECTION AVOIDED! — Likelihood of Being Human-Written: 96%

Test #6: Cover Letter

As mentioned earlier, the next three use cases will each use a different prompt. For the “Cover Letter” use case, I’ll be using this application for a Chief Technology Officer job opening.

And here’s what Originality thinks of the modified output from HideMyAI:

AI SPOTTED! — Likelihood of Being Human-Written: 14%

Test #7: Cover Letter

Next, I’ll use this ChatGPT-generated pomade marketing copy.

Here’s how well Originality.AI stacks up against HideMyAI’s generated output:

DETECTION AVOIDED! — Likelihood of Being Human-Written: 89%

Test #8: Email

To close this out, I asked ChatGPT to write me an email to a college professor asking for an extension.

Again, let’s determine which one between HideMyAI or Originality comes out on top:

DETECTION AVOIDED! — Likelihood of Being Human-Written: 97%

HideMyAI vs. Originality.AI: Different Levels

Besides use cases, you can also tweak the readability levels of HideMyAI’s output. This ranges from:

• Professional — Regular
• Professional — Casual
• Grade School
• Secondary School
• Undergraduate
• Masters
• PhD

Unlike the previous section, I’ll use only one ChatGPT prompt for all tests. I asked ChatGPT to generate a one-paragraph essay about the wheel, and here’s what the chatbot wrote:

To avoid inconsistency, I’ll use the “General Writing” use case for all texts. So, enough chit-chat, let’s begin our last batch of tests.

Test #9: Professional (Regular)

DETECTION AVOIDED! — Likelihood of Being Human-Written: 61%

Test #10: Professional (Casual)

DETECTION AVOIDED! — Likelihood of Being Human-Written: 59%

Test #11: Grade School

DETECTION AVOIDED! — Likelihood of Being Human-Written: 100%

Test #12: Secondary

DETECTION AVOIDED! — Likelihood of Being Human-Written: 100%

Test #13: Undergraduate

DETECTION AVOIDED! — Likelihood of Being Human-Written: 100%

Test #14: Masters

DETECTION AVOIDED! — Likelihood of Being Human-Written: 100%

Test #15: PhD

AI SPOTTED! — Likelihood of Being Human-Written: 1%

The Count

Time for the final tally. The table below provides us with an overview of how well each test went, and its last row shows the average AI-detected content for the 15 modified texts.

After rigorous testing, I found an average likelihood of 22.07% that Originality.AI detects AI writing from tweaked texts from HideMyAI. This is remarkable compared to the testing I’ve done with Undetectable AI and Originality AI, considering that the latter only detected a likelihood of 54.89% AI content from the former’s outputs. 

However, there are some things we must consider, including:

• Variability: Despite my best attempts, these are still isolated cases. You’ll likely get a different result if you perform your own testing. This is only meant to give an overview of what to expect, but not what will exactly happen.
• Readability: I also haven’t touched upon the readability of each modified text because that’s not the main point of this article. However, it’s worth noting that I can’t consider any of the generated HideMyAI texts to be passable as human using my own criteria.
• What’s Next: Today, AI bypass tools are a new invention to answer a market need. AI detectors haven’t had a chance to create solutions for this subset of AI content yet. With enough time, these detectors will surely be able to flag AI-rewritten content more accurately.

For now, I’m ready to declare HideMyAI as the winner of this head-to-head battle.

So, What Now?

HideMyAI is a double-edged sword. This tool can make AI-generated content undetectable but for a steep cost. To better explain, I suggest scrolling up, carefully reading the modified text from earlier, and returning once you’re done. 

All caught up? Good.

The thing with HideMyAI is that you will sacrifice a lot of the readability that ChatGPT, Bard, or Jasper offers to generate undetectable content. More often than not, it also fails to follow the parameters you set for the output. This is because HideMyAI (and Undetectable AI, for that matter) focuses on bypassing AI tools by using slang, simplifying sentences, and (sometimes) creating content with wrong grammar or punctuation. 

As a writer, I’ve always supported modifying AI content using your own words. ChatGPT content is already a great framework for what you’re trying to write — now, make it your own by adding your own flair and supporting information. 

However, if you need something quick and easy, I’d still recommend HideMyAI. But always remember: modify further. Clean up their output and ensure that it fits your target audience.

That’s the correct approach in using HideMyAI to avoid detection tools such as Originality. Good luck!