Figuring out how to avoid AI detection in writing isn’t necessarily about trying to get past AI detectors (although we regularly review the detectability of AI writing tools that attempt to do so by rewriting and rephrasing content).
It’s about striving to create unique, high-quality content in a world where AI slop is impacting the internet, and AI errors, as well as hallucinations, are becoming a noticeable issue (such as the Deloitte AI mistake and the newspaper AI book list scandal).
Remember, generative AI tools and LLMs (large language models) are trained on human work and ideas — they don’t have any original thoughts of their own.
So, if human-written text flags as AI (also known as a false positive), it’s a sign that the words, phrases, and structure may be really similar to what’s already out there.
And how are you supposed to stand out and add value for readers if you’re just recycling what’s already online?
Here, we’re going to cover what AI detection tools are and what they actually do, and give you some tips on how to make text not AI detectable.
Avoid AI detection in writing with these tips:
An AI detector is a tool that analyzes a given text to determine the likelihood of it being generated by AI tools like ChatGPT. Tools like Originality.ai’s AI Checker can tell if you used AI when creating content (such as for generation or editing content).
After analysis, AI detectors typically assign a score indicating how confident they are that the text is likely AI or likely human. Learn more about AI vs. human content detection scores.
The keyword to remember here, though, is likely.
Originality.ai has a high 99% accuracy for AI detection.
However, accuracy rates can vary by product and model, so it’s important to check the accuracy rating of the AI detector you choose.
It’s also important to note that AI detectors can produce false positives, meaning users should always do some further investigation when determining whether a text is AI.
Writers and students can also use tools like Originality.ai’s Chrome Extension with features like ‘Watch a Writer Write’ to showcase the entire writing process.
If you’re interested, you can learn more about how AI content detection works at Originality.ai.
Some of the common issues with AI writing include:
As you’ve probably guessed, keeping these elements out of your writing is an important part of avoiding the question “did AI write this?”
However, they’re not all to keep in mind.
Learning how to make text not AI detectable may just be easier than you think.
Here’s a quick checklist of how to make text sound less like AI:
✓ Don’t let AI write everything; instead, write authentic human-written copy.
✓ Use an AI detector for transparency that includes features like Deep Scan to ethically improve writing.
✓ Switch up your subtopic structure
✓ Change up your word choices and sentence structure
✓ Add some personal touches that add value to content (keep Google’s E-E-A-T in mind)
✓ Avoid robotic or formulaic sections
The best way to avoid AI detection in writing? Write authentic human-written copy.
Using AI as a support (such as for grammar edits) rather than a replacement is the best way to get an original score and pass AI detection, because the copy is just that — original.
One of the best ways to avoid AI detection in writing is to scan it with a tool like Originality.ai’s AI Checker and edit any flagged text (keep reading for details on Deep Scan).
See, tools like Originality.ai aren’t just for editors and publishers — they’re for writers too.
By scanning your article, blog post, or copy before submission, you can see if your copy is Likely AI.
Originality.ai makes finding the Likely AI areas easy too, with its color-coded highlighting.
For example, let’s take a look at some AI-generated text:
Then, let’s run that through the Originality.ai AI Detector:
An Originality.ai AI detection scan uses color-coded highlighting to indicate how confident it is that sections of text are AI-generated. In the image above, we analyzed AI-generated text, which the red shades of corresponding highlighting reflect.
To get here, all you need to do is import or copy and paste your text into Originality.ai’s Content Scanner and click “Scan.”
When an AI detector is confident that it’s human-written, you can feel confident that you’re producing original, useful content…
…and not just recycling the same old ideas and structures that already dominate the search engine results pages (SERPs).
Beyond just checking whether your text is Likely AI, with Originality.ai, you can now also use Deep Scan.
Deep Scan is like an AI-powered editing sidekick that gives you tips and suggestions on how you can authentically improve your writing and make it sound less like AI wrote it.
That way, you can ethically craft high-quality and engaging copy that sounds human and engages your audience and readers.
Learn more about Deep Scan.
If you’ve ever generated a full article from ChatGPT or other generative AI tools, you know that it tends to look smooth and uniform throughout.
Just look at the middle of this AI-generated article:
As a writer, how often do your subtopics come out looking that consistent?
Human writers rarely structure every single subtopic in a piece exactly the same way, whether on purpose or accidentally.
After all, the goal when writing is typically to create something engaging and informative, and not all subtopics benefit from the same kind of structure.
Some may work better with a list thrown in there, while others are better suited for a narrative approach. You may have a personal experience that helps illustrate one idea, and others that would carry more weight with a quote or statistic.
Besides, varying your subtopic structure naturally makes it less predictable, which makes it harder to mistake for AI. Tools like ChatGPT love writing that looks organized and perfect, but it’s just not natural for humans — especially ones concerned with creating authentic, relatable, people-first content.
Generative AI tools love using certain words and phrases, and this hasn’t gone unnoticed. Check out some of the most obvious ChatGPT words and phrases.
The reason why LLMs like ChatGPT tend to repeat certain words or phrases is that they generate text based on patterns learned from their training data.
OpenAI uses this example in their explanation of how ChatGPT is developed:
“For example, during the model’s learning process (known as “training”), the model might be tasked with completing a sentence like “Instead of turning left, she turned _____”. Early in training, its responses are largely random. However, as the model processes and learns from a large volume of text, it becomes better at recognizing patterns and predicting the most likely next word.”
In other words, if ChatGPT has a ton of content marketing articles that start with something like “In a digital world…” in its training, it’s likely to do the same when asked to create an article on the subject.
The same goes for sentence structure — the more they see certain structures in relevant training data, the more likely they are to use and repeat them.
So, to avoid that kind of repetitiveness, try:
Mix short, medium, and long sentences to give writing more of a natural rhythm and make it sound less robotic
Example of repetitive sentence structure: “I think that [statement]. I also think that [related statement].”
Alternative: Vary the sentence structure by using different connecting words or phrases, such as “Furthermore,” “Moreover,” or “In addition,” or rephrase the sentence structure entirely.
If the same word keeps popping up in your work, consider replacing it with a synonym, rephrasing it, or removing it entirely.
Example of repetitive language: “This is a great product. I really like this product. This product is amazing.”
Alternative: Use synonyms, antonyms, or related words to express the same idea without repeating the same words or phrases.
If a common phrase has been around for a long time, there’s a good chance that ChatGPT has come across it during training. Say it in your own words to keep things sounding truly original.
Example of a cliché: “Actions speak louder than words.”
Alternative: Avoid using clichés or common phrases, and try to express the same idea in your own words.
One important note here: just because something looks or sounds like AI doesn’t mean it is.
Multiple studies have found that humans aren’t as good at recognizing AI text as they think.
The good news is that you can provide clarity about a Google Doc’s origins by using a tool like Originality.ai’s Chrome extension.
By enabling this extension, you can show clients or editors a character-by-character replay of how your content came together, proving that you didn’t just copy and paste something from an LLM.
It may be able to write text, but there’s one thing that you can bring to a piece of content that ChatGPT absolutely cannot: real, human experiences. And it just so happens that this aligns perfectly with the E-E-A-T principles Google uses to evaluate page quality:
Google introduced the extra “E” for “Experience” in an update of its search rater guidelines in 2022. This means it considers whether content has been produced by a writer with real, first-hand experience on the topic when determining search engine rankings.
When you put this together with the fact that Google has penalized sites with AI content since its March 2024 update (which significantly reduced low-quality, unoriginal results in search), you can see how important it is to avoid publishing unhelpful, even spammy, machine-written articles that just don’t align with its guidelines.
And a big part of that is including personal experiences in content.
Not only does it demonstrate that you have some kind of firsthand knowledge of what you’re talking about, but it also adds an element of originality to your content.
Besides, it can also make your writing much more engaging for readers. Who wants to read the same old story churned out by ChatGPT over and over again?
Although studies do suggest that humans can’t detect AI text on their own, we can tell when something sounds a little robotic or formulaic.
Just look at the conclusion from this AI-generated article:
Starting the ending with “In conclusion…” followed by “By…” and listing the main points in the article is a classic conclusion structure.
So, the key to avoiding AI detection here (and writing more unique copy for readers) is switching things up.
Stop following the same old patterns, and start trying something new!
Here are a few tips to help you break out of robotic or formulaic patterns in writing:
Avoiding AI detection in writing is becoming key to ensuring that your original, human-written content stands out in a sea of AI slop.
AI detectors may seem like they’re just for publishers and editors, or anyone else worried about the source of content.
However, they can also be extremely helpful for writers who want to elevate their craft and want their voices heard above the mass-produced AI junk that some churn out to try and get ahead.
Want to see the difference it makes in your writing? Try Originality.ai’s AI Checker and its breakthrough Deep Scan feature today!
Learn more about AI detection:
Still wondering about a few things? Check out some of the top FAQs about avoiding AI detection in writing:
There are several things you can do to make your writing not detectable by AI checkers, including creating original, human-written copy.
Then, you can avoid common AI writing issues to make your writing even more engaging for readers as well by writing content in distinct sections, changing up word choices and sentence structures, adding personal experiences, and rewriting robotic-sounding or formulaic text.
AI writing detectors can be hard to trick, especially when they’re 99% accurate, like Originality.ai’s AI Checker. So, instead of trying to trick them, you should focus on creating original writing that doesn’t sound formulaic or robotic, and actually adds value for readers. AI only recycles what’s already out there.
Even our own AI detector is not perfect, and can produce false positives. These false positives can be very painful for anyone who creates original content — whether you’re a student wrongly accused of using ChatGPT by TurnItIn or GPTZero or a writer who received a false positive from Originality.ai.
AI & Plagiarism Detector for Serious Content Publishers
Originality.ai
64 Hurontario St
Collingwood, Ontario
L9Y 2L6
