How to Spot AI Affiliate Slop Before You Click

By Austin Little

You searched for a tool. You got a page that looks like a review, reads like a brochure, and was probably written by a machine that never touched the thing it's ranking. Here's how to tell, in about a minute, before you hand over a click, an email address, or a card.

AI disclosure. This page was drafted with AI assistance and edited for Apiary. We don't invent quotes, stats, people, or events.

Yes, an AI-assisted article about AI-generated slop. I know. The difference I care about is whether a human with something at stake checked the claims, said plainly how the page was made, and refused to make things up. That's the standard I'll ask you to hold other pages to, and you should hold this one to it too.

Search for "best free AI writing tool," "best transcription app," or "best AI for small business," and you'll see a familiar shape: a long page with a numbered list, a comparison table, star ratings, pros and cons, and a big colorful button next to each pick. Some of these pages are written by people who tested the tools. Many aren't. A growing share look like they were generated in bulk, stuffed with affiliate links, and published to catch your search.

I'm not going to name and shame specific sites. Sites change, owners change, and I'm not in a position to prove who wrote what. What I can do is teach the tells, so you can judge any page yourself, including this one.

What "affiliate slop" means

Let's separate three things that get lumped together.

Affiliate links. A link that pays the publisher a commission if you buy through it. That's not inherently bad. Lots of honest reviewers earn money this way. Google's own spam policy says plainly that not every site in an affiliate program is a thin affiliate, and that good affiliate pages add value with things like original product reviews, rigorous testing and ratings, and product comparisons.

AI-assisted writing. Using a language model to help draft. Also not inherently bad. This article was drafted with AI help, and I've told you so at the top.

Affiliate slop. Pages produced mainly to rank in search and collect commissions, with little or no original testing, often generated at scale, often padded, and frequently wrong. The writing might be AI, might be a content mill, might be both. The problem isn't the tool. It's the absence of anyone who actually used the product and cares if you get burned.

Google describes related behavior in its spam policies. It defines scaled content abuse as many pages generated mainly to manipulate search rankings rather than to help users, and lists "using generative AI tools or other similar tools to generate many pages without adding value for users" as an example. It describes thin affiliation as affiliate content where product descriptions and reviews are copied from the merchant without original content or added value, often appearing as "cookie-cutter sites or templates with the same or similar content replicated" across domains.

Those are Google's definitions for its own search ranking, not law. But they're useful language for what we're looking at.

Why this matters more for AI tools

Tool roundups for AI products are a particularly good breeding ground for slop:

• The products change constantly. Features, model names, free tiers, and prices shift every few months. A page written once and never updated drifts into fiction.
• "Free" is slippery. Many AI products offer a free plan, a free trial, or "free credits" that require a card. A careless page flattens those into "free."
• It's hard to test well. A real review of a writing tool means using it on real work for a while. That's slow and expensive. Generating a list is fast and cheap.
• Readers are often new. If you're searching "best AI tool for X," you may not know enough yet to spot a wrong claim. That's exactly who slop is aimed at.

The cost to you isn't just a wasted click. It can be a card on file for a trial you forget about, an account that keeps your uploaded documents, or a tool that doesn't do the thing you needed.

The one-minute scan

Before you read closely, scroll the page once, top to bottom, and look for these. If three or more show up, treat the page as advertising until proven otherwise.

1. No named author, or a name with nothing behind it.
2. No date, or a date that looks auto-updated but the content clearly isn't.
3. Every product gets four stars or more.
4. Pros and cons that could apply to any product.
5. No screenshots of actual use, or only the vendor's marketing images.
6. Prices or plan details stated with no "as of" date.
7. A "Buy now" or "Try free" button next to every item, and no disclosure near it.
8. The same structure repeated for every pick, down to sentence length.
9. No mention of free or open-source alternatives that don't pay commissions.
10. A conclusion that says "the best tool depends on your needs."

None of these alone proves anything. Together, they tell you whose interests the page serves.

The tells, in detail

Tell 1: Nobody actually used it

The single biggest tell is the absence of evidence that a human used the product.

Real testing leaves fingerprints:

• Specific, slightly inconvenient details. "The export button is buried in a menu under the share icon." "It choked on a 40-minute file and I had to split it."
• Screenshots of the reviewer's own work, not the vendor's polished demo.
• A clear description of what they tested it on. "I ran three blog drafts and one meeting transcript through each."
• Things that went wrong.

Slop tends to describe products using the vendor's own feature list, rephrased. If every description sounds like the product's homepage, that's probably where it came from.

The FTC's endorsement guidance has a line worth remembering here. Among the core principles for endorsers, it says: you can't talk about your experience with a product if you haven't tried it. A page that implies hands-on experience without any is, at minimum, misleading you about what kind of page it is.

Tell 2: Generic pros and cons

Read the pros and cons for three different products. Could you swap them without anyone noticing?

Slop pros: "User-friendly interface." "Powerful features." "Great for beginners and professionals alike."

Slop cons: "Can be pricey for some users." "Learning curve for advanced features." "Free plan has limitations."

These are true of almost every software product ever made. They tell you nothing. A real con is specific: "The free plan caps uploads in a way that makes it useless for anything over a short clip." A real pro is specific: "It was the only one that kept speaker labels straight with three people talking."

Tell 3: Stale or undated facts

AI tools change constantly. A page that states a price, a model name, a word limit, or a free-tier allowance without saying when it checked is a page you shouldn't trust on those details.

Look for:

• "As of [month, year]" next to prices and limits.
• A visible "last updated" date that matches the content. If the page says "updated this month" but describes features that you know were renamed or removed, the date is decoration.
• Model names that are generations behind.

The rule I use: any price or limit on a third-party page is a rumor until I see it on the vendor's own pricing page. Always click through to the source before trusting a number.

Tell 4: "Free" that needs a card

This one costs people real money. Watch for the blur between:

• Free plan: you can use it indefinitely, with limits, no payment method.
• Free trial: you can use it for a while, then it stops or converts to paid.
• Free credits with a card: you must add a payment method to get the "free" part.

A slop page will often say "free" for all three. An honest page will say which one it is, and whether a card is required at signup.

Before you sign up for anything a roundup called "free," go to the vendor's pricing page and the signup flow. If it asks for a card, stop and decide whether you're okay with that. Set a calendar reminder before any trial ends.

Tell 5: The disclosure is missing, vague, or buried

The FTC's staff guidance on endorsements says affiliate marketers should disclose their relationship to the retailer clearly and conspicuously, and that the closer the disclosure is to the recommendation, the better. It gives a sample: "I get commissions for purchases made through links in this post."

The same guidance says some wordings aren't enough on their own. It says consumers might not understand that "affiliate link" by itself means the person placing the link is getting paid, that a "buy now" button would not be adequate disclosure, and that "commissionable link" is probably not clear. It says "paid link" right next to an affiliate link should be adequate. It also says a single disclosure on a site's home page isn't sufficient, and that a hyperlinked "DISCLOSURE" button isn't enough because many people won't click it.

That's the FTC's staff guidance for advertisers and endorsers, and the FTC itself says the guidance isn't a safe harbor and that context matters. I'm not telling you any given page is breaking the law.

What to look for as a reader:

• A plain sentence near the top or right next to the links saying the publisher earns money from purchases.
• Wording that names the relationship. "We earn a commission" beats "This post may contain affiliate links."
• No disclosure at all on a page with a button next to every product? Assume they're getting paid and you're not being told.

Tell 6: Everything wins

If a "best of" list has ten items and all ten get 4.3 stars or higher, the ratings aren't measuring anything. Real testing produces spread. Some tools are bad. Some are good at one thing and bad at another. An honest reviewer will say "skip this one."

Watch also for "best for" labels that give every product a trophy: "Best overall," "Best for beginners," "Best for teams," "Best value," "Best for power users." Sometimes that's useful. Often it's a way to avoid ranking anything below anything else, so every link stays clickable.

Tell 7: The free and open-source options are missing

Here's a tell specific to AI tools. Many useful jobs can be done with free, local, open-source software: local language models through Ollama, local transcription with Whisper or whisper.cpp, and so on. Those projects typically don't pay affiliate commissions.

If a roundup of "best AI tools for X" lists only paid products, or only products with affiliate programs, and never mentions that a free local option exists, ask why. Sometimes it's because the reviewer doesn't know. Sometimes it's because there's no commission in it.

This isn't proof of bad faith. But an honest guide to "best" should at least mention the free path, even if it concludes the paid tool is worth it.

Tell 8: Machine-shaped prose

AI-generated text has a texture. It's getting harder to spot, and these signs are hints, not proof. People write this way too, and good AI-assisted writing can avoid all of them. But clusters are telling:

• Throat-clearing intros. "In today's fast-paced digital landscape, choosing the right tool can be overwhelming."
• Perfectly balanced sections. Every product gets the same number of paragraphs, the same number of bullets, the same sentence rhythm.
• Hedge-everything conclusions. "Ultimately, the best choice depends on your unique needs and preferences."
• Overused vocabulary. Words like "seamless," "robust," "game-changer," "elevate," "unlock," "streamline," piling up.
• Confident claims with no source. "Studies show users save hours per week" with no link to any study.
• Weird errors in specifics. A feature attributed to the wrong product, a model name that doesn't exist, a price for a plan the company doesn't offer. Language models produce fluent wrong details.

That last one is the most dangerous. Fluency isn't accuracy. A sentence can be grammatically perfect and factually invented.

Tell 9: Invented testing numbers

Watch for precise-sounding results with no method: "Tool A was 37% more accurate." "We measured 2.3x faster output." Ask: measured how? On what? Compared to what? Who ran it?

Real benchmarks come with a method you could, in principle, repeat. If the page gives a number but no method, treat the number as decoration.

Tell 10: The page doesn't help you decide

The real test of any review is whether you're better able to choose after reading it. A good one tells you who should pick what and why, and who should skip it. It warns you about the gotchas.

Slop leaves you exactly where you started, plus ten links. Its job was to get the click, not to help.

A worked example (fictional)

Here's a made-up excerpt, built from common patterns. It's not copied from any real site.

3. ScribeFlow AI — Best for Content Creators ⭐ 4.7/5 ScribeFlow AI is a powerful, user-friendly AI writing assistant that helps you create high-quality content in seconds. With its intuitive interface and robust feature set, it's perfect for bloggers, marketers, and businesses alike. Pros: Easy to use. Powerful AI. Great templates. Cons: Free plan is limited. Can be pricey. Pricing: Free plan available. Paid plans start at affordable rates. [Try ScribeFlow Free →]

Let's run the tells:

• No evidence of use. Nothing specific, nothing went wrong, no screenshots.
• Generic pros and cons. Swap the name and it describes any writing tool.
• "Free plan available," but no detail on what it includes or whether a card is needed.
• "Affordable rates" instead of a price and date.
• A "Try free" button with no disclosure next to it.
• High rating with no method.
• Vocabulary: "powerful," "user-friendly," "intuitive," "robust," "perfect for... alike."

Now an honest version of the same slot:

3. [Tool name] — good for short drafts, weak on long ones I used it for two weeks on blog intros and one 2,000-word draft. On intros, it was fine: fast, readable, needed light edits. On the long draft it lost track of the outline after the third section and repeated a paragraph. Exporting to Markdown dropped my headings. Free plan (checked [date] on the vendor's pricing page): [what it includes]. Signup did / did not ask for a card. Skip it if: you write long pieces or need clean Markdown. We earn a commission if you buy through this link.

You don't need to know anything about the tool to see which one did the work.

What to do instead of trusting the list

Go to the source

For every tool you're considering:

1. Read the vendor's own pricing page. Note the date. Note whether there's a free plan, a free trial, or a card-required credit.
2. Read the vendor's own docs for the feature you care about. Does it actually do the thing?
3. Start the signup flow and see what it asks for before you commit. You can back out.
4. Read the privacy policy or data page if you'll upload anything personal. Look for whether your content is retained or used for training.

Look for independent signs of real use

• Forum threads or community discussions where people describe actual problems. These are messy, but real complaints are hard to fake at scale.
• The project's own issue tracker if it's open source. You'll see what breaks.
• Reviews from people with a track record you can check: a name, past work, a history of saying "don't buy this."

Try the free path first

For many AI jobs, you can find out whether you need a paid tool by trying a free or local one first. Apiary has guides on using AI without paying and on running local models. If the free option does the job, you've saved the money and kept your data. If it doesn't, you now know exactly what you need a paid tool to do, which makes you much harder to sell to.

Use your own small test

Before committing to any tool, give it one real job from your own work, the same job you'd give every candidate. Compare results side by side. Ten minutes of your own testing beats ten pages of someone else's star ratings.

The reader's checklist

Print it, screenshot it, stick it on the monitor.

Before you click a "Try free" button:

• [ ] Is there a named author with a real track record?
• [ ] Is there a clear, nearby disclosure of how the page makes money?
• [ ] Does the review show evidence of actual use (specifics, screenshots, things that went wrong)?
• [ ] Are prices and limits dated, and do they match the vendor's own page today?
• [ ] Does "free" mean free plan, trial, or card-required credits?
• [ ] Does any product get a "skip it" or a real criticism?
• [ ] Are free or open-source alternatives mentioned?
• [ ] Do the numbers come with a method?

Before you sign up:

• [ ] Did the signup ask for a card? If yes, is that okay with you?
• [ ] Did you set a reminder before any trial ends?
• [ ] Do you know what the service does with what you upload?

When you've been burned

If a "free" trial charged you, or a tool wasn't what a review promised:

• Cancel the subscription through the vendor's account settings, and screenshot the confirmation.
• Contact your card issuer if you were charged for something you believe you didn't agree to. They can explain your options.
• Report deceptive practices. In the US, the FTC takes reports at ReportFraud.ftc.gov. Its endorsement guidance page points there too.
• Report spammy search results.

For publishers: how not to become the problem

If you run a site, including a small one, here's the short version of how to do affiliate content without becoming slop.

• Use the thing. If you haven't used it, don't review it. Say "we haven't tested this" if you mention it.
• Disclose plainly, near the links. The FTC staff guidance gives simple wordings that work, like "I get commissions for purchases made through links in this post."
• Date your facts. "Checked [date]" next to prices and limits.
• Say "skip it" when it's true. Your credibility is the only asset you have.
• Mention the free path, even when there's no commission in it.
• Disclose AI assistance. If a model drafted it, say so, and say what a human checked.
• Don't scale what you can't check. Google's spam policies call out generating many pages with AI "without adding value for users" as scaled content abuse. More pages isn't the goal. Better pages is.

Apiary doesn't run affiliate links on these craft guides.

Short answers

Are all affiliate review sites bad? No. Google's own spam policy says good affiliate pages add value with original reviews, rigorous testing, and real comparisons. The problem is pages that do none of that.

Is AI-written content automatically slop? No. The question is whether a human with something at stake checked it, disclosed how it was made, and refused to make things up.

How do I know if a "free" tool really needs a card? Go to the vendor's pricing page and start the signup. Don't trust the roundup's word for it.

What's the fastest single tell? No evidence that anyone used the product. If you can't find one specific, inconvenient detail, the "review" is probably a rephrased brochure.

Where do I report a deceptive review in the US? ReportFraud.ftc.gov. For spammy search results, Google's spam policy page links to its search quality user report.

The bottom line

You don't need to be an expert to spot slop. You need one minute and a little suspicion. Look for a real author, a plain disclosure, evidence of use, dated facts, and at least one honest "skip it." Check every price and "free" claim on the vendor's own page. Try the free path first. And remember the page's job, if it's slop: to get your click. Your job is to get what you came for without getting burned.

References

• FTC — FTC's Endorsement Guides: What People Are Asking — https://www.ftc.gov/business-guidance/resources/ftcs-endorsement-guides-what-people-are-asking (fetched 2026-10-01)
• Google Search Central — Spam policies for Google web search — https://developers.google.com/search/docs/essentials/spam-policies (fetched 2026-10-01)
• FTC ReportFraud — https://reportfraud.ftc.gov/ (fetched 2026-09-25)