How to cheat AI Content Detectors

As it stands, copywriters already have a few tools at their disposal to cheat AI Content Detectors, but their clients are the ones who ultimately place the orders.

AI Content Detector Tools could be better; they’re at the time of writing (January 2023, things in AI develop rapidly) unreliable.

The point is that cheating AI detection techniques is laughably simple.

Universities, teachers, content managers in corporations, and SEO Agencies all have a legitimate interest in learning how they might be tricked. And there is Google, who also does have a legitimate interest in not showing the same spammy content over and over again. They must understand how writers might fool their content quality control and the AI Content Detection systems.

Some strategies are already shared in Reddit, private internet marketing groups or given away as a “How to make 7 figures with ChatGPT in 7 days” Twitter Secrets, others are SEO approaches that have been around for over many years now.

Let us dig deeper on some of the more popular methods people use to fool AI Content Detectors.

How can you trick an AI Content Detector?

There are a few simple techniques to cheat AI Content Detectors into thinking a piece of text was produced by a human.

AI Cheat 1: Combine multiple prompts to make long-form content

The first and most obvious way to trick an AI Content Detector is to use a longer form of content that is made by combining multiple prompts. Nearly all the AI writer tools, like Jasper.ai, Copy.ai, and Writesonic, have a tool called “Blog Wizard”,”AI Writer”, “Magic Blog Writer” or something similar.

Outline and Article created entirely by WriteSonic AI

All of them require you to make an outline and a few headings for an article first. Then, based on the headings, the tool will make the whole article. Each heading is an instruction for the AI to make a paragraph. Then, these paragraphs are put together to generate the final article.

It will be difficult for the AI Content Detector to tell that the article was created by an AI if you just copy and paste an average article of 1000 words written in response to several prompts.

Let’s say you want to see if such an article was “AI-Generated” by pasting it into an AI Detector tool like for example Huggyface OpenAI Detector.

If you are lucky and can reach the page (it is constantly overcrowded lately) you will very likely obtain a human-generated grade since the models at the base level aren’t designed to recognize such broad strokes of artificial intelligence. Not that they will always fail, but the odds are stacked against success.

AI Cheat 2: Rephrasing and rewording

It is possible to trick AI Detection by changing the language and phraseology of one’s sentences. Even the internal AI Watermarking prototype that OpenAI is building may be beaten with enough effort, as Scott Aaronson – an OpenAI engineer – confirmed in his blog:

Now, this can all be defeated with enough effort. For example, if you used another AI to paraphrase GPT’s output—well okay, we’re not going to be able to detect that. On the other hand, if you just insert or delete a few words here and there, or rearrange the order of some sentences, the watermarking signal will still be there. Because it depends only on a sum over n-grams, it’s robust against those sorts of interventions.

Obviously, paraphrase may be done with varying degrees of effort.

There has also always been a multitude of easy web-based tools available to content agencies, SEOs, and copywriters for paraphrasing, rewriting, and generally making the material “unique enough” to pass such detection tools. There are many out there, we present only a few of the more popular ones here:

Paraphrasing via WriteSonic, Jasper.AI

SEOs have been using content rewriting tools for over several years to avoid the dreaded duplicate content “penalty” (wherein their material is demoted in search engine results because it is too similar to that of other websites).

WriteSonic Rephrasing Content

WriteSonic or Jasper AI are not only AI Copywriting tools but also powerful paraphrasing and rephrasing tools that use algorithms to automatically rewrite text. They can be useful for avoiding plagiarism, simplifying complex ideas, summarizing lengthy texts, and even expanding your writing skills. To use them, simply copy and paste the text you want to paraphrase into the tool, it will generate a list of paraphrased versions for you to choose from.

Paraphrasing via QuillBot

A further tool to assist rewrite and paraphrase text is QuillBot. It works as an extension for Microsoft Word or in a web browser and has a powerful website editor, too.

Quillbot is a tool that can help you paraphrase text. It uses artificial intelligence to understand the meaning of the original text and generate a new version that conveys the same information in different words. You even can change the tone of the paraphrased text, shorten, simplify or expand the original text.

Using Quillbot is easy. You simply input the text you want to paraphrase and the tool will generate a new version for you to use. You can then review the paraphrased text and make any necessary edits to ensure that it accurately conveys the meaning of the original.

Grammarly

Grammarly is helpful for both rewriting and paraphrasing.
Grammarly originally only detected orthographic and grammatical errors, but in the last year it was constantly enhanced.

You may now use Grammarly to

• change active to passive and vice versa
• “rewrite for clarity.” ( 😉 )
• and much more text changes with just one click.

These kinds of modifications are performed by artificial intelligence algorithms included into Grammarly.

Can non modified ChatGPT Text identified by Google & Co?

Conclusion

As you can see, learning how to cheat AI content detectors is not particularly difficult. With the right tools and techniques, it’s possible to create content that is very hard for AI algorithms to detect as being originally generated.

This can be useful if you want to pass off AI-generated content as being written by a human, or if you simply want to create unique, high-quality content without having to spend hours manually writing and editing. However, it’s important to note that intentionally attempting to deceive AI content detectors may be unethical or even illegal in some cases, so it’s important to use these tools and techniques with caution and consideration for their potential consequences. (The last two sentences were by the way written by ChatGPT, funny, isn’t it?!).