AI News HubLIVE
In-site rewrite2 min read

AI system spots fake reviews with 93% accuracy on Amazon, 91% on Yelp

Online shoppers could one day face fewer misleading fake reviews thanks to a newly tested AI-powered detection system developed by researchers at the University of East London.

May 21, 2026

AI system spots fake reviews with 93% accuracy on Amazon, 91% on Yelp

by University of East London

edited by Sadie Harley, reviewed by Robert Egan

Sadie Harley

Scientific Editor

Robert Egan

Associate Editor

Editors' notes

This article has been reviewed according to Science X's editorial process and policies. Editors have highlighted the following attributes while ensuring the content's credibility:

fact-checked

trusted source

proofread

The GIST

Add as preferred source

Credit: Pixabay/CC0 Public Domain

Online shoppers could one day face fewer misleading fake reviews thanks to a newly tested AI-powered detection system developed by researchers at the University of East London.

Fake reviews are a growing problem for consumers and online businesses, especially with the growth in AI-generated content. According to researchers from the Royal Docks School of Business and Law, misleading reviews can distort competition, damage trust in online marketplaces and persuade people to buy poor-quality or even unsafe products.

The new system combines AI language analysis with behavioral clues such as whether the emotional tone of a review matches its star rating, how long the review is and other patterns linked to suspicious activity. The researchers, from the Royal Docks School of Business and Law, say this gives the model a fuller picture of whether a review is genuine or deceptive.

The new study, published in FinTech and Sustainable Innovation, describes a new "hybrid fusion" model designed to identify fraudulent reviews on platforms such as Amazon and Yelp.

Unlike older systems that mainly relied on keywords or simple patterns, the new approach is designed to understand the meaning and context behind written reviews. That helps it detect more convincing fake reviews that might otherwise appear genuine to shoppers.

In testing, the model achieved 93% accuracy on Amazon review data and 91% accuracy on Yelp reviews, outperforming several traditional detection methods examined in the study.

Co-author Dr. Hisham AbouGrad said, "Fake reviews are becoming increasingly sophisticated and harder to detect. Our findings show that combining AI language understanding with behavioral signals can provide a more reliable way to identify misleading reviews and help strengthen trust in online marketplaces."

Co-author Fiza Riaz said, "This research shows that AI systems can move beyond simply spotting suspicious words. By looking at context and behavior together, the model can better recognize patterns linked to deceptive reviews while still supporting genuine customer feedback."

The paper says the next stage of the research will focus on improving the system using larger and more varied datasets, exploring newer AI models and studying how the technology could eventually work in real-time on large e-commerce platforms.

More information

Hisham AbouGrad et al, Metadata-Enhanced Hybrid Fusion Architecture: Commercial Fake Reviews Detection Model Using Transformer Embeddings, FinTech and Sustainable Innovation (2026). DOI: 10.47852/bonviewfsi62028859

Key concepts

Large language modelsAI alignmentHuman-centered AI interfacesTrustworthy machine learningGenerative AI misinformation

Provided by University of East London

Citation: AI system spots fake reviews with 93% accuracy on Amazon, 91% on Yelp (2026, May 21) retrieved 22 May 2026 from https://techxplore.com/news/2026-05-ai-fake-accuracy-amazon-yelp.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.

Explore further

Detecting fraudulent product reviews with enhanced accuracy

6 shares

Facebook

Twitter

Email

Feedback to editors