Last month, a friend recommended a restaurant in Southbank based on glowing Google reviews. When I looked more closely at those reviews, something felt off. The language was too polished, too consistent across different reviewers. The phrasing patterns were almost identical. The reviews read like they’d been written by the same person — or the same machine.

I started digging. What I found was a growing problem that’s affecting how Australians choose where to eat.

The scale of the problem

AI-generated restaurant reviews are everywhere. A 2025 analysis by digital marketing researchers estimated that between 10 and 20 percent of new restaurant reviews on major platforms are now AI-generated. In competitive dining precincts — CBD areas of Sydney, Melbourne, and Brisbane — that number might be higher.

The reviews come from a few sources. Some restaurants commission them directly, using services that generate fake positive reviews. Some come from reputation management firms that create “supplementary” reviews for clients. And some are generated by bots as part of competitive attacks — negative AI reviews aimed at rival restaurants.

How to spot them

After reading hundreds of reviews with suspicious patterns, here are the tells:

Generic descriptors. AI-generated reviews love phrases like “a symphony of flavours,” “impeccable presentation,” and “a delightful culinary experience.” Real humans rarely write like this. They say things like “the pasta was good but the bread was stale.”

No specifics. Fake reviews tend to describe the restaurant in general terms without mentioning specific dishes, prices, or details that only someone who actually visited would know. If a review doesn’t mention a single dish by name, be suspicious.

Perfect grammar, zero personality. Real reviews have typos, weird capitalisation, and personal tangents. AI reviews are grammatically flawless and personality-free.

Review clusters. If a restaurant suddenly gets fifteen five-star reviews in a week after months of nothing, that’s a pattern worth questioning.

Profile checking. Click on the reviewer’s profile. If they’ve reviewed thirty restaurants in five different cities in the past month, all with five stars and similar language, it’s not a real person.

Why this matters

For restaurants playing it straight, fake reviews create an unfair competitive landscape. A new restaurant doing genuine work can be buried by a competitor that purchased fifty AI-generated five-star reviews.

For consumers, the erosion of trust in reviews means we lose a useful information source. I used to rely on Google reviews for quick assessments of new places. Now I discount them heavily and go back to asking friends, reading professional reviews, or just walking in and taking my chances.

The platforms are trying to fight back. Google has ramped up its detection algorithms and removed millions of suspicious reviews in 2025. Some hospitality businesses are working with AI consultants in Melbourne to develop their own review monitoring and response systems. But the AI generation tools are improving just as fast. It’s an arms race with no clear winner.

The restaurant perspective

I talked to three restaurant owners in Melbourne about this. Two admitted they’d been approached by companies offering AI review services. Both declined, but one said the temptation was real — especially when competitors seemed to be benefiting from artificially inflated ratings.

The third owner had experienced the other side: a sudden cluster of negative reviews that appeared to be AI-generated, likely from a competitor. Getting those removed from Google took weeks of back-and-forth with the platform and cost the restaurant real business in the meantime.

What platforms should do

Better detection is necessary but not sufficient. Platforms need to:

• Require verified visit data before allowing reviews (some are starting to do this)
• Weight reviews from established, verified accounts more heavily
• Provide restaurants with better tools to flag and contest suspicious reviews
• Be transparent about their detection methods and removal rates

What you can do

Read reviews critically. Look for the specifics. Check the reviewer’s history. Pay more attention to detailed negative reviews (which are harder to fake) than glowing positive ones. And when you genuinely enjoy a meal somewhere, leave an honest review. The real ones matter more than ever.

The trust problem in online reviews isn’t limited to food, obviously. But food is personal. Where we choose to eat matters to us. And when that choice is being manipulated by machines, we should care.