Google has developed a new machine learning algorithm to fight more effectively against fake reviews that pollute Search and Maps. In 2023, over 170 million fraudulent reviews have been removed!
On Google Maps, reviews often allow users to learn about different places and points of interest before going there. Are the dishes at this restaurant good? Are the rooms in this hotel good? Is this museum interesting? So many questions that users’ comments can answer. In 2023, there were around 20 million contributions per day on Maps and Search, counting business hours updates, ratings, photos, reviews, videos… The problem is that comments and reviews “customers” are far from reliable. And it’s not about to get better, with generative AIs like ChatGPT made available to the general public. Indeed, they make it easy to generate text on demand in a natural language. The perfect tool for writing fake reviews across the chain – and not just positive ones, they can also be used to criticize competing establishments. This is why Google has developed a new machine learning algorithm, which allows the American giant to detect fake reviews on Search and Maps more quickly and more efficiently, as explained this blog post.
Fake Google reviews: more than 170 million fake reviews detected
In 2023, this new algorithm allowed the company to remove 45% more fake reviews than the previous year, for a total of more than 170 million fake reviews. Business owners were also protected from more than 2 million attempts by bad actors to claim business profiles that didn’t belong to them – an increase of more than 1 million per year. compared to 2022. More than 12 million fake business profiles have also been removed or blocked.
If reviews are rigorously controlled by Google’s moderation teams, and “its devices constantly monitor contributions to detect any suspicious activity”this new algorithm examines daily the “longer term signals”, which improves its ability to identify both isolated cases, such as users posting the same review on different listings, and larger-scale fraudulent campaigns, such as a business receiving a sudden spike in 1 or 5 star reviews . Google indicates, for example, that it has identified a network of scammers inviting users to post multiple reviews on specific listings and to click on online advertisements in exchange for imaginary remuneration. “In just a few weeks, we detected more than five million fake review attempts linked to this scam,” explains the Internet giant. This is called being effective!