Amazon Harnesses AI to Combat Fake Customer Reviews

Amazon checks reviews before putting them online. They use smart computer programs (AI) to look for signs that a review might be fake. Most reviews pass the test and are posted quickly, but some are carefully checked by Amazon to make sure they’re real and honest.

“If Amazon is sure that a review is fake, they act fast to stop it or delete it. They also take additional steps if needed, like stopping the person from writing reviews, blocking accounts of dishonest users, and sometimes even going to court against those involved,” the company explained.

Amazon uses advanced technology to analyse various data to identify fake or incentivised reviews. Machine learning models examine seller activities, customer reports, behavioral patterns, review history, and more.

Large language models and natural language processing are employed to detect anomalies that suggest a review might be fake or influenced by incentives like gift cards or free products.

Deep graph neural networks are utilised to understand complex relationships and behaviors, helping identify and remove groups of dishonest users.

Josh Meek, a senior data science manager at Amazon, highlights the challenges in distinguishing between authentic and fake reviews, citing examples like rapid review accumulation due to advertising or a customer suspecting a review based on grammar.

To enhance review integrity, Amazon also brings in expert investigators when needed. In 2022, the company blocked over 200 million suspected fake reviews globally, demonstrating its commitment to maintaining a trustworthy review system in its online stores.