More on News
While Amazon is working on various strategies to ensure the authenticity of customer reviews, media reports have highlighted methods for identifying fake reviews on the platform. The surge in artificial intelligence (AI) technology has led numerous companies to implement effective AI-powered solutions. Amazon, too, is leveraging AI to combat fake reviews.
According to Amazon, when a customer submits a review, the company’s AI solution analyzes to identify known indicators of fake reviews before publication. Although most reviews meet Amazon’s rigorous authenticity standards and get posted immediately, some undergo additional company scrutiny.
If Amazon identifies a review as potentially fake, the company quickly blocks or removes the review. In more severe cases, additional measures are implemented, like revoking a customer’s review permissions, blocking accounts engaged in deceptive practices, and taking legal action against the involved parties,” the company stated.
Amazon explains that its machine learning models analyze proprietary data. This includes the seller’s advertising investments, potential influences on increased reviews, customer-reported abuse, risky behavior patterns, review history, and more.
The integration of large language models (LLMs) and natural language processing techniques allows to work in tandem with data anomalies, helping to detect signs of fake reviews or incentivized using a gift card, free product, or other form of reimbursement.
The company also highlights using deep graph neural networks to analyze and comprehend intricate relationships and behavioral patterns to identify and remove groups involved in deceptive practices.
“The difference between an authentic and fake review isn’t always apparent to someone outside of Amazon,” said Josh Meek, senior data science manager on Amazon’s Fraud Abuse and Prevention team.
“For instance, a product may accrue reviews rapidly due to a seller’s advertising efforts or offering a high-quality product at a competitive price. Or a customer may think a review is fake based on factors like poor grammar,” Meek added.
Amazon further engages in expert investigators if a review is suspicious or requires additional evidence. In 2022, Amazon blocked over 200 million suspected fake reviews across its global stores.