Google shared a report explaining how it is detecting and removing fraudulent content utilizing brand-new machine learning models.
Google announced a major update to their machine learning algorithms last year, which made it much easier for Google to spot new abuse trends than in years before.
For instance, despite there being millions of profiles, Google's automatic systems discovered a dramatic increase in Business Profiles with websites ending in .design or top.
To prevent false or fraudulent content from being uploaded to the Google Maps system, Google's systems check new content before it is posted.
In some areas, scammers have begun adding false phone numbers over user-contributed images to deceive naïve consumers into calling the scammer rather than the legitimate company.
Google explains that to avoid this issue they implemented a new machine learning algorithm that is capable of identifying numbers superimposed on shared images.
According to Google, they are currently scanning images to identify any superimposed content aimed at redirecting phone calls from a company to the scammer's phone number.
Their approach involves verifying the absence of bots, identifying replicated content, detecting resemblances to established fabricated reviews through specific word patterns, and leveraging a system called "intelligent text matching" to recognize misleading content.