AI-generated reviews. 

Harmless nuisance? Or the harbinger of a shifting landscape that will rewrite the way businesses and customers interact in a way that we can’t even begin to anticipate?

In either case, they’re here and they’re proliferating. 

So let’s talk about it. 

What’s Even Real Anymore?

AI-generated reviews are growing at a pretty hot clip. 

According to TripAdvisor, 10.7% of reviews on their site were AI-generated in 2024. That’s more than double the number of AI-generated reviews they reported in 2023.¹

It’s the same story in ecommerce, with Temu reporting that 10.9% of the reviews on their site are AI in 2025, up from 5.17% in 2024.²

So we know that more and more online reviews are coming from large language models like ChatGPT and Claude. 

Unfortunately, they’re also getting increasingly difficult to spot. Sometimes, it’s comically easy, like this classic example. Even if they had cut out that telltale first line, it still includes phrases like “creamy and full of flavor, a perfect start to the meal.” People don’t talk like that. 

Here’s another one. I would bet the family ranch that this is AI-generated. 

See how it says a lot but also says nothing at all? That’s AI, baby. 

But some AI-generated reviews can be tricky to identify, especially if you’re not looking for them. 

One study by researcher Balázs Kovács at Yale School of Management showed that both people and AI detectors struggle to differentiate between AI and human-written reviews. The vast majority of people asked to differentiate between AI-generated and human-written reviews couldn’t do so consistently.³ 

And it’s only going to get harder. AI can include colloquialisms and slang, intentional spelling and grammar errors, and “personal” details when asked to. Perfect grammar and em-dashes* won’t be tells for long. 

*If you want to know when I turned on AI, it was the moment I heard that the em-dash—that’s this guy—was being considered an “AI tell.” It’s my favorite punctuation. How. Dare. They.

Not All AI-Generated Reviews Are Created Equal

I see two main categories of AI reviews, and they’re not at all the same thing. 

The first category is an AI-generated review created at an individual’s request based on a real experience. A guest comes to your hotel or restaurant and decides to leave a review. So they pop the details into ChatGPT and ask it to whip up a review based on their experience. 

That’s what we saw in the first example above. There are enough real details (I had the linguini, hubs had the patty melt, Tori and Ashley were our servers) to make me confident it’s based on a real meal. 

The second category is capital-F-Fake, created wholesale by a bot and based on nothing at all. These are spam, possibly set up by the business themselves to boost their rating. This is illegal, as the FTC has banned this kind of fake review.⁴ 

This is what we saw in the second example above. No dishes, no drinks, no servers, not a single specific detail. Fakety-fake. 

The first type? Not a biggie, in my view. Yes, it would be nice if the reviewer wrote it themselves, but writing isn’t everyone’s strength. As long as the review accurately describes the experience, no real harm done.  

The second type? A biggie. Fake reviews erode trust in all reviews. How should we believe any review when half of them are fake?

But therein lies the problem. As a consumer and a writer, I can often clock AI-generated reviews. But I can’t always tell if a review is the first type (based on a real experience) or the second type (completely fake). They both use the same LLMs. They both have the same writing style. And they both may get flagged by AI detection software. 

What All This Means For You

For Business Owners

Problem the first: If you’re not using fake reviews but your competitors are, you could be at a disadvantage. If the hotel up the street is sitting pretty at 4.9 stars thanks to fake reviews, it could cost you bookings and business. 

Problem the second: Even if every review on your site is customer-submitted, you could be falsely accused of cheating with fake reviews if those customer-submitted reviews are AI-generated. Remember, the language can be very similar between the customer-submitted and bot-submitted reviews.

For the Industry

Problem the first: I’m very concerned about the loss of consumer trust in reviews in general. Currently, about 75% of people trust online reviews.⁵ But some already believe that companies pay to bury bad reviews and even pay for higher star ratings. I think we can expect to see that trust degrade as AI-generated reviews proliferate further. 

Problem the second: AI-detection software is in an arms race with LLMs.

Attitudes toward AI are complex. Some find it to be a useful tool for research and specific problem-solving. I do, em-dash complaints notwithstanding. But AI-generated content feels inauthentic and gives a lot of people the ick. So LLMs have a vested interest in making their models sound more and more “human.” 

As AI models release new versions that can “write” more organically, the detectors release new versions to catch up. It’s going to get harder, not easier, to determine what’s human and what’s machine.

Problem the third: For review-driven industries like hospitality, a loss of faith in reviews could have a real impact on your bottom line. How will purchasers make decisions if they can’t trust user experiences? Will they be less likely to take risks on new restaurants or hotels if they don’t feel confident in what they read online?

Navigating the New Review Minefield (Without Exploding)

It’s complicated out here, y’all. But here are some suggestions. 

Don’t Post Fake Reviews

Just don’t do it. Illegal. Unethical. Bad news all around. 

Guest Verification

If you collect reviews natively on your own site, a “verified purchaser” or “verified customer” tag on reviews can go a long way to building trust. Many of the larger ecommerce sites like Amazon have this feature. 

If you collect your own reviews, there are third-party platforms that can help you with review verification, like this one. (Not an endorsement, I’ve never used them.) 

Monitor and Flag Reviews

You should be keeping an eye on your reviews anyway, since they’re one of the best tools you have for guest feedback. But in addition to spotting unhappy guests and important takeaways, you can also look for AI content. 

Look for: 

• Telltale intros that people forget to delete. “Here’s a review you could use…” or “As an AI language model…”
• Lots of superlatives (impressive, perfect, memorable) but not a lot of detail. What did they order, who was their server, what was the space like? What was the occasion, who did they go with, and what didn’t they like?
• An overly formal tone that sounds more like a marketer who’s trying too hard than a real person. 
• A formulaic mix of review topics, again without specifics. Service, ambiance, food, menu selection, value.
• Yes, SOMETIMES an excessive number of em dashes. But an em dash on its own isn’t a red flag! 

When you spot it, flag it. Review platforms do actively take down a lot of AI content, but they can certainly use our help in finding it.

Request Reviews at Your Own Risk

It’s super common to send post-stay or post-purchase emails or SMS texts asking for a review. Others request reviews in person with a QR code on a card in guest rooms or presented with the check.

If you’re collecting reviews for your own website, that’s fine. But if you’re relying on Google or Yelp or Tripadvisor, you need to be more careful with requesting reviews. 

Yelp, for example, has a policy against requesting reviews. They will hide any review they deem solicited in “Not recommended” purgatory where they don’t count toward your star rating.⁶ Google, on the other hand, straight up encourages you to request reviews and offers tips on how to do so.⁷ 

So make sure you know the solicitation rules for your preferred review platforms. 

Whether you request reviews or not, be wary of incentivizing reviews. It’s very tempting to offer a discount or freebie to get more customer reviews. But it’s against the terms of use for most review sites, and can get your legit reviews flagged and taken down. If it’s for your own website, you can offer an incentive, but you can’t condition it on being a good review. And you have to disclose the incentive. 

Encourage Video/Photo Proof

I have a hunch that the average joe can spot an AI-generated video more easily than AI-generated text. Ill-defined hands, uncanny valley skin, and a mouth that doesn’t quite match up with the words being spoken tell the brain that something is off, even if your eye isn’t quite sure what it is. 

So encourage photos and videos if it’s allowed—see above about requesting reviews. People (read: me) often forget that they can share multimedia, so give ‘em a little reminder.

Looking Ahead

Review platforms have a vested interest in minimizing fake content, and they’re always rolling out new detection models. Far beyond just assessing the text, they look for things like clusters of reviews from a single IP address or posted in a short amount of time. 

I’ll be interested to see if review sites implement verified stay or purchase requirements as well. While these are common on large ecommerce sites, you don’t see them on major platforms like Google or Yelp. But as fake content proliferates, I wouldn’t be surprised if verification rolls out in the near future. 

The best way to get real, quality reviews? Be awesome at what you do. Which obviously, you already are.

1. Fake AI TripAdvisor Reviews Increased by 137% from 2019 to 2024, Originality.AI, February 10, 2025 ↩︎
2. 1361% Increase in AI Temu Reviews From 2022 to 2025, Originality.AI, June 24, 2025 ↩︎
3. The New Food Critic: AI-Generated Restaurant Reviews Fool the Best of Us; Yale School of Management, August 29, 2024 ↩︎
4. Federal Trade Commission Announces Final Rule Banning Fake Reviews and Testimonials, FTC.gov, August 14, 2024 ↩︎
5. Online Review Trends and Statistics, Pissed Consumer, January 2024 ↩︎
6. Don’t Ask For Reviews: Why Yelp Does Not Recommend Solicited Reviews, Yelp, January 2017 ↩︎
7. Tips to get more reviews; Google Business Profile Help ↩︎