There is a new kind of junk on websites, and it does not look like spam.
The comment is polite. The grammar is clean. Sometimes it even mentions the topic of the post. There is no “viagra”, no casino link, no obvious scam. And still, after two sentences, you can tell that nobody actually read the article. That is AI slop — the shape AI comment spam usually takes on a website.
What AI slop is
AI slop is mass-produced text that sounds fluent and contains almost nothing: no experience, no fact, no opinion that belongs to a real person. It is not “using ChatGPT to draft a paragraph”. It is the leftover: empty comments, generic reviews, form messages that could be pasted onto any site in any language.
On a website it usually shows up in the places you built for real people:
- Comments that start with “Great insights, thanks for sharing!” and go nowhere
- Reviews made of five adjectives and no product detail
- Contact-form messages like “I wanted to know your price” in a language nobody on the site uses
- Guest posts that restate the headline for 800 words
Why this is harder than classic spam
Classic spam is noisy. You see the link, you see the keyword, you hit Trash. AI slop is quiet. It passes a CAPTCHA because a person, or a well-built bot, submitted it. It passes a word blacklist because it never uses the words you blocked. It can even be on-topic.
The cost is different too. You do not lose an afternoon to casino links. You lose it deciding whether a comment is a real reader or generated filler. Real discussion dries up because the thread already looks full. Search engines still treat that empty text as user-generated content on your page.
This is not the same problem as AI crawlers reading your site. Those bots take content out. Slop puts content in. We already wrote about AI-generated comments as a spam type in wpDiscuz. This post is the other half: what you are actually seeing, and what you do with it.
The awkward case
Here is the question that does not have a clean answer. An AI-written comment with no link, no promo, and a vaguely relevant sentence: is that spam?
Some site owners publish it. Some delete it on sight, because it trains readers to ignore the comment section. Some ask the person to rewrite.
Our working position is this: an empty comment with no experience and no facts does not belong on the page, even when it is polite. We get why some owners publish it anyway — it looks harmless and deleting it can feel harsh. If you disagree, say so in the survey. That is the question we actually want answered.
A short survey
Four questions. Answer in a comment, like this: 1B, 2A, 3C, 4D. If you can, paste one real example that made you stop and think “is this a person?”.
We will collect the answers and publish the results in a separate post.
1. Do you see AI slop on your site?
- A. Regularly
- B. Sometimes
- C. I am not sure if it is AI or a real visitor
- D. Not that I have noticed
2. Where does it show up most?
- A. Comments
- B. Reviews / testimonials
- C. Contact forms
- D. Registrations
- E. Guest posts
3. What is harder to moderate now?
- A. Classic link spam
- B. Fluent AI-looking text
- C. Both, about equally
- D. I have not noticed a change
4. An AI comment with no link, but on topic. What do you do?
- A. Publish it
- B. Delete it
- C. Ask the author to rewrite
- D. Depends on the case
If the queue is already full
If this is already eating your moderation time, the useful check is not “does this text sound human?”. Human-sounding is the point of the generator. What still works is who sent it, from where, and how the form was submitted. That is the layer we run in CleanTalk Anti-Spam: reputation and behavior, not a grammar test.
The Real Person badge
If you do not want to judge every comment by hand, there is a signal for verified authors. The Real Person badge marks the author of a comment or review who passed Anti-Spam checks and confirmed their email. It is optional and off by default. The how-to is in that post.
We will collect the answers and publish the results in a separate post.
Leave a Reply