Copy article

Fool-proof trick to fil­ter out the bogus reviews from the real

ended 27. May 2026

Research has found about half of reviews left on 30 of the biggest review plat­forms are fake.

This is based on ana­lysis by tech firm TruthEn­gine of mil­lions of reviews and cus­tomer rat­ings left on plat­forms such as Amazon, Tri­pad­visor, Trust­pi­lot and Google.

Since April last year it has been illegal to leave a fake review. But mil­lions are still being writ­ten for Brit­ish busi­nesses – in many cases by the com­pan­ies them­selves as they try to win cus­tom.

TruthEngine says an almost fool-proof trick to fil­ter out the bogus from the real is to read only the two- and three-star reviews. Fakers typ­ic­ally won’t bother with these, par­tic­u­larly three stars as it doesn’t move the dial in either a pos­it­ive or neg­at­ive dir­ec­tion. So they are very likely to be genuine.

  • Any other tips for spotting bogus reviews?
  • How worried should we be of AI spitting out thousands upon thousands of fake reviews?
  • What does half of reviews being fake mean for trust in products and services?

Responses by tomorrow.

4 responses from the Newspage community

Copy all

Copy

Editor
Mitali Deypurkaystha
AI Governance Strategist & CMO at AISensa
An arXiv study found people can only identify AI-generated fake reviews with 50.8% accuracy, which is basically the same as guessing with a coin toss, and that should terrify online shoppers. Reviews used to help people avoid wasting money, but AI is turning them into industrial-scale persuasion machines where thousands of believable fake opinions can be generated in minutes. Reading two- and three-star reviews is still one of the smartest tricks because they tend to sound more balanced and useful, but once businesses realise consumers trust those more, AI can mass-produce “believable middling reviews” too. The safest approach now is to treat glowing ratings with a pinch of salt and pay more attention to things like refund policies, free trials and whether a company is confident enough to let customers walk away, because five stars no longer automatically mean five-star quality.
Copy

Editor
Nouran Moustafa
Practice Principal & IFA at Roxton Wealth
Fake reviews are dangerous because they do not just mislead people; they distort trust. Consumers are already overwhelmed by choice, and reviews are often used as a shortcut when people do not have time, confidence or technical knowledge to properly compare products and services.

My tip would be to look for balance. Real reviews usually have texture: what went well, what could improve, and specific detail about the customer journey. Fake reviews often sound either too perfect, too generic, or oddly dramatic. I would also check the timing. A sudden flood of five-star reviews in a short period should raise eyebrows.

AI makes this much more worrying because fake reviews can now be produced at scale and made to sound more human. That means platforms need stronger verification, not just better wording filters.

If half of reviews are fake, the bigger issue is that consumers may stop trusting reviews altogether. That hurts good businesses as much as customers.
Copy

Editor
Tony Redondo
Founder at Cosmos Currency Exchange
Online reviews on open networks where anyone can post leave them vulnerable to drive-by manipulation, with research showing half of the reviews across major platforms like Amazon, Google, and Trustpilot are fake. To filter out the bogus on an open platform, look to two- and three-star reviews; fakers rarely bother with lukewarm ratings that don't move the statistical dial. Be wary of flawless syntax, a telltale sign of modern AI-generated feedback, alongside single-review accounts and sudden bursts of activity. This industrial-scale manipulation fuels a dangerous feedback loop where AI models absorb fake reviews as truth, reshaping search results and breaking consumer trust. Better still, use an evolved platform like Trustist that acts as a closed aggregator, collecting verified post-purchase feedback and pulling in data from multiple sites to build a secure, fraud-resistant organic digital footprint.
Copy

Colette Mason
AI Ethics Consultant at Clever Clogs AI
Old tricks like copy-paste review farms and the same sentence appearing word-for-word across dozens of profiles or sites, have largely disappeared. AI killed that. Every generated review is now linguistically unique but look for the substance being similar in terms of scope or structure.

Real reviews mention the dodgy zipper, the weird smell out of the box, the fact it took three weeks to arrive. Fakes are enthusiastic about everything and specific about nothing.

Check the timing. Real satisfaction trickles in over weeks and months. If hundreds of five-star reviews appear within days of a product launch, that's a clue it's a campaign. Spikes are evidence, because the AI can't fake the calendar.

Check the reviewer, not just the review. No profile picture, no review history, or a history that spans wildly unrelated products with camping gear, dental supplies, industrial adhesive, all reviewed within the same week? That's someone being paid by the review, not by the experience.