You finally landed a freelance writing client, and you’re ecstatic.
You go through the onboarding process with your new client, and they give you a content brief in the next few days.
So you get to work. You write the content and source examples and use Grammarly to check for errors.
You go the extra mile and get your friend to proofread it over and finalize your article. You send your project to your client and wait for feedback.
They emailed you back and can’t wait to hear how they liked your article.
The email says they ran your article through an AI detection tool, and it came out over 90% AI-generated.
Because of this, they no longer want to work with you.
In one of my posts about freelance writing and AI challenges, I did not foresee clients accusing writers of using AI writing tools.
This happened to one of my Writeto1k students, and I’ve heard similar stories all over Twitter.
I’m frustrated and concerned about clients not giving their writers the benefit of the doubt and relying on faulty AI detection tools to check our writing.
If a client accuses you of using AI to write your post (and you didn’t use AI for writing), I’ll share how to AI-proof your writing and what to do about those types of clients.
What Is AI-Generated Content?
OpenAI is the company that made ChatGPT and AI-generated text popular and changed the content landscape forever.
When a person or writer prompts an AI writing tool like Jasper AI or ChatGPT to create something, the pre-generated content is the output.
For example, here’s a prompt from Jasper AI on interview questions.
While a lot of pre-generated content can be highly valuable, as a freelance writer, you will probably see this type of content as:
- Creating misinformation
- Possibly plagiarized
- Lacking originality
- Not in the first person
- Unable to produce thought leadership content
But many freelancers use AI writing tools. I do!
AI’s advantages help us speed up productivity, avoid writer’s block, and give us new angles.
I’ve been using ChatGPT and Jasper for short-form content like a Tweet, a paragraph for a sub-topic, email subject lines, and for ideas.
How to Detect AI Generated Content (What Clients Are Doing With Your Article)
Since the explosion of ChatGPT, OpenAI has created an AI detection tool.
This tool will tell you if the content is likely AI or not.
Other AI detection tools were born, like Originality AI, Content at Scale, Writer, and GPTZero.
Because Originality AI is a paid tool, businesses feel this would be the best AI generator content checker for them. But Originality is highly inaccurate.
I’ve plugged in Jasper AI content, and it said it was 100% human.
I’ve put in 100% human content (a blog post I wrote years ago), and it came out as overwhelmingly AI written.
But it’s not just Originality giving you false positives. All AI-detection tools are inaccurate.
OpenAI has a detection tool, and even they say it’s not reliable.
“In our evaluations on a “challenge set” of English texts, our classifier correctly identifies 26% of AI-written text (true positives) as “likely AI-written,” while incorrectly labeling human-written text as AI-written 9% of the time (false positives). “
This is NOT a perfect science, but clients are using these tools as the end all be all to their content plan and accusing writers.
This is wrong.
Poor writers and mediocre writers are everywhere and there’s a massive problem for clients to find good reliable writers. Because of this they want to ensure their content is up to their standards and for many of them, and that means relying on AI detection tools.
Six Ways to AI-Proof Your Writing
But this isn’t a problem of clients plugging in our post into an AI-detection tool.
It’s a writing problem, and as writers, we need to do MORE than just spit out a 1,000-word blog post.
Here are my recommendations for AI-proofing your writing while improving your writing at the same time.
1. Create a Fleshed Out Outline (And Have Your Client Approve It)
If your client provides content briefs, then you can give a fleshed-out outline for approval.
This goes for clients that give you the outline. GoDaddy provides the outline and I flesh it out.
I use Google Docs to create my outline, and send it to the editor.
They review my outline and add notes (you can see where they highlighted my topics).
To flesh out my outline, I provide subtopic examples by mentioning what I will be doing.
So, for the three-step approach to Instagram influencer marketing, I share that visuals are good for unboxing videos and say I can show screenshots on my iPhone using InShot.
I’m going the extra mile for this client and for my article and will gladly share that in my outline.
For my other subtopics, I write out the definitions of what I mean crystal clear messaging means and give specific examples I will reference in my post.
Adding these extra elements to my outline shows my client that these ideas are MINE and aren’t generated by an AI tool (although you can use it for ideation and getting a rough draft of your topics. I do this with an SEO tool).
2. Snazz Up Your Writing With SMEs and Examples
If you aren’t landing freelance writing jobs and not gaining some amazing testimonials for your writing, you might be creating mediocre content.
This is content that anyone can write. The topics are also, “meh,” and have been created thousands of times (think, “how to lose weight,” “easy meal planning tips”).
And you know what? This is the type of writing you get from AI writing tools.
For example, here is what ChatGPT gave me for an outline about losing weight in your 40’s.
This is generic, and everyone knows these tips.
It’s mediocre writing.
There are no, “latest discoveries” about weight management and losing weight.
There is no subject matter expert (SME) quotes from doctors or dieticians about the fat epidemic in the US.
And no actionable ideas or examples to help a person in their 40s start a weight loss plan.
Having examples and SMEs peppered throughout your content makes it valuable, original, and AI-proof.
In my masterclass Writing a B2B $1000 Article, I share my proven tactics to gain a lot of money with my content, and part of those tactics is securing SMEs for my content.
Lately, I’ve been using Twitter to source original content.
I will poll my audience to share in articles.
This gives me a snapshot of a topic I want to know about.
But, if I want to quote SMEs, I will ask a question and use the replies for quotes.
Using some of these replies instantly makes my content much more valuable than AI written content.
And one way to turn your mediocre writing into high-paying writing is to describe and talk about your quotes or examples.
I may screenshot a Tweet and, in my article, explain what it is and WHY it’s important for my article.
Those two things – SMEs and original content – make my content highly authentic and humanly written.
But what happens if the client comes back to you and STILL stays your content didn’t pass an AI-detection tool?
Well, tell them that you spent time crafting a pitch for HARO and hours going through the quotes and finding quotes on LinkedIn and Twitter.
Let them know you took that information and expertly weaved them into your content, and all of this should help your clients see that you are writing authentically.
And if you ARE using AI to help with your rough draft? What do you say?
Tell them you used AI and that you optimized it so that it reads like something you wrote.
3. Use Several AI-Detection Tools
Copy and paste your article into different AI detectors. Give your client all those scores when you submit your article.
Why? Because some AI-detectors will say that it’s mostly AI, others will say it’s primarily human-written, and others will say it’s 50-50.
Here’s the result of a final draft of a client piece in GPTZero.
This tool says my content is entirely human-written. Now, let’s put that same piece into Originality AI.
Unfortunately, Originality AI can only do 1,000 words, while GPTZero took my entire post of 3,000 words.
This tool says my content is 54% AI.
Now let’s look at Content At Scale’s tool.
This tool says my content is 90% human written.
So, you see?
Different results mean AI-detectors aren’t reliable and even account for false positives.
4. Use Grammarly Last
Prior to all of this talk about AI writers, I would write the content, self-edit, and then use Grammarly.
But not anymore. I have to check my self-edited content first with AI-detectors.
Because Grammarly is an AI writer, the AI detectors will flag it if I run my 100% humanly-written content through Grammarly.
If you use any editing tool, your content will ALWAYS have some AI writing (because you agree with their suggestions).
Tell your clients that you use Grammarly or another editor, which will account for some AI in your post.
And make sure and also check a plagiarizing tool like Copyscape and then tell your client the results.
The next two points weren’t mentioned in my video.
5. Use the Google Doc History
Now – I’ll be the first to say that while my clients pay me over $1,000 to write for them, I wouldn’t want them to view my writing process.
Because it’s crazy and sloppy, I write and re-write, take days and erase and come back and change things ALL.THE.TIME.
But Google Docs, has a history tab. Go to File> Version history.
And, because of this, I use different docs for each phase of my writing.
- One doc for a dump file (add all my links and examples)
- One doc for my outline
- One doc for a rough draft
- One doc for continuing the rough draft
- One doc for editing
- One doc for proofreading
- One doc for Grammarly
- One NEW doc for the client with completed and edited post
So, you see, a client would get no history from me because I did all the work on a different document.
But not anymore, I’m afraid.
I don’t like, as you can see, how long it takes me to write my content; sometimes, I do not touch it for days!
Oh well, if that is what clients want to view to ensure it was humanly written then so be it.
6. Don’t Write Like a Robot!
The last suggestion is a hard one for me.
I’ve been writing online since 2014, and I know for a fact I write like a robot.
And since we are using OUR writing to train the robots, it’s no wonder OUR writing reads somewhat robotically.
But it’s not 100% of my writing.
As you saw my scores for my client piece, the most it was deemed AI was 54%.
But, if you’re getting results like 97% AI for many tools, you could be writing like a robot.
Robots like to be wordy and passive, so you’ll improve your scores if you can work on those two things alone.
Let’s look at my finished draft for GoDaddy.
This is what I wrote:
The highlighted sentence is a mouthful, and I didn’t realize it when I submitted my post.
But the GoDaddy editor did and magically changed it.
The editor removed “advancement and achievements” to “leaps” and added emphasis to “is.”
Using this format shortened my piece to make it more clear.
Another thing AI writing does is write in the third person. AI content has no personalization.
For example, Jasper or ChatGPT might say,
“Businesses know that email goes hand in hand with their marketing efforts for more leads.”
If you’re writing a blog post and the audience is businesses, then you can write in the second person.
“You know that email goes hand in hand with your marketing efforts, but it can also help with brand messaging.”
The second example, is personal, as if I’m talking to my audience.
One last thing to improve your writing is to be clear in every sentence you write.
Use as few words as possible to make your point and break up your writing as I do for my blog post.
Your goal is to get your reader to read past the headline, the first sentence, the second sentence, and so on.
What Does Google Say About AI-Generated Content?
One reason why clients are using AI-detection tools is that they think Google won’t rank their content.
And in the past, Google has said AI generated content is against guidelines.
But, for today, this isn’t true.
Liaison to Google, Danny Sullivan, has said that AI content isn’t bad.
And I know many niche website owners exclusively use AI content to rank in Google.
Educate your clients on this!
What To Do If You Are Accused?
I wouldn’t work for that client anymore. I would say that I have many happy clients that pay for my content, and if they feel that way, we should part ways.
Be civil and move on. Why would you want a client who checks your content with an unreliable tool?
And, even if you do all six of these things and your client STILL doesn’t believe you, then move on…you don’t need that headache and worry second guessing your writing.
I’ve worked with many clients, and I have dumped many clients too.
This is YOUR business, and you do what YOU want!
Let me know in the comments if you had a client say you used AI to write your post when you never did!