AI Policy & Enforcement

The Policy

Some competitors have blurred the lines arguing that you as a client just care about quality regardless of who (or what) is involved in delivering that final output. This is the excuse they use to “enhance” their process with AI writing workflows, often without informing their clients.

While it’s true that clients typically look for quality above all, this is not relevant for the sake of crafting this policy, in our opinion. Our stance here is that at ContentPit if you’ve paid for human-written quality content, that’s what we intend to deliver. You’re not paying for “quality content written by anyone or anything”.

If the service is not clearly labeled “human” or “AI”, then it’s safe to assume that it can include a mix of both and that we don’t have specific restrictions on the use of AI for that particular service. This is often the case with services that render content that is not going to be used in “high-risk environments” such as Google organic search, where many people are wary of Google’s stance on AI writing. Example: social media services.

Note that we allow using SEO tools, grammar-checking tools…etc that have AI features, as long as none of the tools is used to write anything in the final document for our “human written” services.

Our internal policy is clear: For our “human written” services, every word in the final document must be written by you (the writer). That’s where we draw the line. This moves helpful AI SEO tools and other tools away from the “line of fire”. Anything labeled “100% human” means the final delivered output has to be crafted 100% by a human.

Enforcement

A little while ago we created an internal SOP to help our editors better understand our AI policy and more importantly, enable them to enforce it. We’ll outline what we do below but please note that we’re intentionally being vague about some aspects to avoid having our processes “reversed-engineered” by any bad actors to come up with workarounds aimed at circumventing enforcement systems.

Our enforcement workflow consists of 5 pillars:

1. Writing workflow monitoring and audit.
2. AI Detection tools (3 major ones and we test new ones all the time).
3. Common sense (we train editors to look for dead AI giveaways such as repeating nouns and sentence structures, rarely using contractions..etc. Just small stuff that often means nothing but can add up when combining them with other “pillars”).
4. Confidentiality. Our writers are not aware of our specific enforcement processes which increases the risk they take if they try to circumvent them.
5. Zero-tolerance. These are tough economic times and we’ve always taken care of our writers even during the toughest COVID days. For years they’ve had a very dependable income stream. We’ve communicated that using AI writing for “human written” services is a zero-tolerance offense if caught and proven. Most writers would not risk everything in an attempt to circumvent enforcement, especially having no idea what the odds are for them to succeed. We’ve also communicated that we will not be issuing warnings if we’re “suspicious”. Instead, their writing will be put under more scrutiny in the background and they’d wake up one day with a termination email in their inbox.

An important note about AI detection tools:

We’ll be 100% honest and upfront here. None of the stuff above is 100% foolproof. In fact, the worst of all from our testing is the AI detection tools. Saying they’re highly inaccurate is an understatement, and we’re not going to name names here but we’ve pretty much tried everything under the sun.

We gain absolutely nothing from tearing them down. It’s actually in our very best interest for these tools to be accurate. That would put AI accusations from clients to bed once and for all, similar to how Copyscape protects us from any plagiarism-related accusations.

So nothing against them, we’re working with one of those major tools to test it thoroughly and are in constant contact with their team for feedback.

While most of these tools have massive potential, they’re really not there yet in my opinion (and with the constant AI advances, may never actually be). Content that was 100% written by me was scoring terribly while I was able to “bypass detection” and get a solid human score by testing one of the more advanced AI writing tools on them.

Please, please take results from these tools with a grain of salt and don’t treat them like Copyscape. We really wish they were the “holy grail” of combating those pesky machines, but they aren’t.

If you wish to discuss specific test results and share your own, please reach out. We’ll be happy to help.

Recommendations

My top recommendation would be to focus on writing briefs that you’re sure would lead to a top-quality article if followed. My second recommendation if you don’t trust us on the AI thing (or just as an additional quality control measure even if you trust us) is to have a VA or an editor skim through the articles to detect any “low-quality” content. 

I advise against asking them to “tell me if you think this is written by AI”. This typically sends them on a wild goose chase, suspecting anything and everything. Since it’s almost impossible to prove 100% beyond a shadow of a doubt whether certain content is AI-written or not, I’d recommend having them focus on the properties of low-quality content like fluff, drifting completely off-topic, not serving the reader’s intent…etc.

If you catch any of this and point it out, you’re protected by our unlimited revisions policy so we’ll work on it till you’re satisfied with the output. If multiple “low-quality content” complaints come from a specific writer we then take them out of rotation and put them on an improvement plan where we monitor them closely. If things don’t improve then they’re let go. If at any point it’s proven that they’ve used AI on non-AI client work, they’re immediately terminated as per our no-tolerance policy.

You’ll always be protected!

It’s important to remember that we’re here to work with you, not against you. We’re not trying to employ tricks for a quick buck. When we use AI writing, we’re transparent about it, as is the case with all of our AI-assisted services. With these services, we make sure we give you a ton of value, and most importantly, our pricing will always reflect the use of AI so you can take advantage of additional savings.

In any case, you’ll always be protected by our unlimited revisions policy and 100% money-back guarantee. I know the “rise of AI” has made us all paranoid. If you’re not convinced we’re being truthful despite everything, ask for your money back. No hard feelings.

Our take on AI content creation and the future

We’re not really about burying our heads in the sand and shouting “Humans will always win!”, “AI will never catch up!”. We’re always looking for ways to give you more value and we’re keeping a very close eye on the AI landscape.

AI content in the SEO world in particular has been very polarizing with multiple camps formed. Some people swear by it and some people hate it. Some people swear by it “when heavily edited by humans” and others wholeheartedly hate it no matter what kind of editing was involved.

The choice is yours. We will always aim to serve the market and your needs. Our job, however, is to make sure you get what you pay for from us. No slimy business, no beating around the bush, no aiming for a 500% profit margin by selling you AI content and passing it off as “human-written”.

What happens when AI replaces what we do, you ask?

Well, the time we’d then save during writing, we’d use to give you more value in other areas. More focus on SEO, more focus on auditing your website and helping you update content, kill non-performing content, help with overall content strategy…etc. We’ve now started offering new services related to social media, newsletters…etc.

We will always aim to provide value and save you time, AI will just help us be faster and better, which means more value at a cheaper price, and we pass on those savings to you. Everyone wins!