Help Wanted: AI Vigilante Task Force

I was on the fence about whether I should write this piece or not. But I decided to do it. I don't want it to be an angry tirade and I am aiming to keep my emotions in check, but it would be dishonest to say I’m not upset.

If you’ve been active on Vocal over the last year you have probably experienced some frustrations with “creators” or “commenters” using AI. I certainly have. And recently there’s been an instance of an account using AI that really floored me in how long it has managed to not only fly under the radar but also receive several awards for AI generated pieces.

I am going to reference the recent example that I encountered throughout the article, but I will leave the user unnamed as Vocal's investigation is still underway. I messaged the Vocal team and ask if I should sit on this piece and wait until they finished looking into the user's pieces, but I was encouraged to go ahead and proceed.

Ultimately my goal is to share some concerns I’ve experienced and also provide some tools for others to navigate this tricky terrain of AI content.

And to be clear I am not writing this piece as a criticism of Vocal or its Team Members. This is uncharted territory and they are doing their best to deal with an absolute deluge of AI generated content. And every time I’ve submitted formal requests regarding issues of plagiarism or AI they’ve always been addressed. Vocal has made their stance on AI very clear: if you’re going to publish AI generated content you need to include a disclaimer. That’s the gist of it, but you can check out the full details in the resource linked below:

But as we all know, a lot of accounts using AI to generate their content don’t want people to know it’s been written by AI. Some of them want to engage in the community and pull the wool over people’s eyes. Why? Simple answer. The same answer for why a great many people want to deceive others: money. But wait. You’re thinking “How would you earn money with AI content?” Well let’s dive into the recent example I mentioned.

This particular user is a very active commenter. And although a lot of their comments have AI generated remarks, they usually do add enough personal phrases to make it seem authentic. And consequently they have a lot of engagement with the comments they leave on pieces.

This account posted a comment on one of my pieces several months ago and I thought it sounded a little formulaic, so I ran the comment through an AI scanner. Even with just a few words it was flagged for AI. I went to their page, opened up their most recent story and ran it through a scanner. It was marked as 100% AI. I clicked the report button on the story and designated it as undisclosed AI (I think Vocal gets way too many of these to really tackle them all).

But since that day I’ve seen this account receive Vocal prizes and get shout outs from other creators for their supportive comments. This surprised me and I decided to take a deeper dive into scanning the user’s pieces. Maybe they’d stopped publishing AI pieces and were getting awarded for their own work?

Nope.

Many of their own published pieces wouldn’t immediately strike you as being AI generated because of the unique formatting choices and a few unique phrases that have been layered in. They publish just enough human or ambiguous pieces that when they slip one in that’s entirely AI, no one raises an eyebrow. And they’ve slipped in more than one…

So over the last few months this account has received not one, but six Top Stories. Several of these Top Stories have a Highly Confident rating on being 100% AI just using a basic scanner. The user's last Top Story has over 50 comments and 70 likes. And according to scans, it's AI generated content. Not only has this account garnered Top Story Awards but they’ve placed on several Leaderboards.

According to my findings they have received the following awards:

• 6 Top Stories ($5 each)
• 2 Honorable Mentions ($3 each)
• 3 Most Discussed Stories - 2nd Place ($10 each)
• 1 Most Supportive Commenters - 3rd Place ($5)
• 1 Emerging Creators - 1st Place ($20)
• 1 Emerging Creators - 2nd Place ($10)
• 1 Emerging Creators - 3rd Place ($5)
• 1 Most Subscribed to Creators - 3rd Place ($5)

That’s a total of $111 in Vocal prizes.

And yeah, this really bothers me. But what can I do about it? None of us joined Vocal to be AI fighter vigilantes. I wish I’d launched a formal complaint instead of just reporting one story all those months ago, but I was busy and didn’t think this user would reach the status that they have.

But maybe if more people had caught on and this user hadn’t gotten traction this wouldn’t have happened. That’s why I’d like to share a few annoying, but hopefully simple steps other creators can implement. That way others can determine if the user that's commenting on their pieces or publishing the pieces they are spending their limited time reading is actually writing them.

If you’re like me, when a user comments on one of my pieces for the first time with something more than a “You’re awesome! Subscribe to me!”, I like to do them the courtesy of checking out one of their pieces. Now if the comment seemed a little formulaic I proceed with skepticism, but even if it seemed authentic I still usually follow these steps:

1. Open one of their recent stories
2. Copy the text from the piece
3. Paste it into a free AI scanner (I use GPTZero, you don't need to make an account at all to run a few scans)
4. Scan the text for AI
5. Check out the results (GPTZero will give you a confidence rating on the result and a probability percentage of the text being AI. It will also highlight AI material in yellow, human created content in green, and leave ambiguous material un-highlighted)
6. If it comes back confident as 100% AI, report it
7. If it comes back as inconclusive, try again on another one of their stories
8. If it come back confident as 100% human, read it
9. If you see an AI user gaining traction, run a few of their pieces through the scanner, download the result reports, and attach them in a formal request to Vocal via the Support page

Nothing’s ever a foolproof process, but I’ve found this to be a quick way of making sure I’m not wasting any time with undisclosed AI content.

You can check comments for AI too, but most free scanners need at least 100 words of text to be 100% conclusive, so you might have to paste a few comments from the same user to check those.

If you've checked a user's work for AI and it came back as human content, but you notice another piece seems a little formulaic, check it! The user I've referenced has several stories that aren't flagged as containing AI content even though the majority are. So give the benefit of the doubt but don't just keep reading blindly if you have concerns.

And one more huge red flag is if an account publishes multiple pieces a day every day. Most real people just can’t generate that much content. It wasn’t the case in this particular instance, but it often is.

I’m not thrilled that this user received Top Stories and Leaderboard placements that someone else deserved. It makes me pretty angry. I hate that there are so many people out there trying to hoodwink us. If someone has a genuine interest in participating in the writing community I want to welcome them with open arms. But I’d much rather read your developing writing with all its flaws than what your AI tool can write for you.

If someone is going to publish AI content, they should include the disclaimer that Vocal requires. If they don't they're not abiding by community guidelines. They should receive consequences and certainly not earn Top Story and Leaderboard prizes.

Now, do I want to spend my Vocal time running AI scans and whistle-blowing? Absolutely not! But I love the Vocal community and the Vocal Team, so I’ll keep trying to do my part to keep it a community and platform that recognizes writers, poets, and storytellers who compose their own pieces. Maybe some of you will be inspired to join me.

Fellow Vocal Creator, Judey Kalchik, has some great pieces on navigating the AI minefield. So if you’re interested in digging deeper, please check out her very helpful insights: