Amazon used machine learning-based detection to block 800,000 fraudulent selling accounts

In its third annual Brand Protection Report, Amazon revealed that it had stopped over 800,000 bad actors’ attempts to set up new selling accounts, with the overall number down from six million in 2020 or more than 50% year over year.

Last year, the tech giant invested over $1.2 billion in technology and experts, hiring more than 15,000 people – including machine learning scientists, software developers and expert investigators – to counter fraud.

Thanks to these efforts, Amazon sued or referred for investigation over 1,300 criminals in the US, UK, EU and China – up from 600 in 2021.

Amazon’s seller verification process requires partners to provide a government-issued photo ID, taxpayer information and details about their identity, location, bank accounts, credit cards, and more. The systems then analyse collected data to detect risk and establish relationships with previously identified bad actors.

Similarly, the firm’s automated technology scans over eight billion listing daily and uses the data to improve its counterfeit prevention tech. The system also relies on customer complaints to enhance the detection process.

“It’s great to see Amazon respond to the threat of counterfeiting and take action against illegitimate sellers, including by supporting us to seize £170,000 worth of suspected counterfeit CDs during raids last year,” Detective Sergeant Andrew Masterson, from the Police Intellectual Property Crime Unit (PIPCU) at City of London Police, said.

The firm also identified, seized and appropriately disposed of over six million counterfeit products in 2022 thanks to its ongoing cooperation with law enforcement.

Subscribe to Charged Retail for free

Click here to get the latest retail tech news free in your inbox each day

“Through our continued investment in advanced machine learning techniques, we have improved our proactive controls, automating and scaling our intellectual property protection and counterfeit detection systems,” Amazon said in a report.

“These systems operate continuously throughout every step of the process—from the moment someone tries to register a new selling account, create a listing, and update a product listing. This includes incorporating the feedback we get from customers, brands, and others.”

In 2022, the number of valid notices of infringement plummeted 35% compared to last year.

According to Dharmesh Mehta, Amazon’s Vice President of Worldwide Selling Partner Services, the ultimate aim is to reduce counterfeits to zero.

Additionally, the firm’s partnership with the United States Patent and Trademark Office (USPTO) allowed it to identify over 5,000 false brands and prevent them from joining Amazon Brand Registry.

Earlier in March, Amazon announced it would start flagging frequently returned items in a bid to help shoppers make more conscious purchasing choices and reduce returns.

Artificial IntelligenceAutomationBig TechNews


Leave a Reply

Your email address will not be published. Required fields are marked *

Fill out this field
Fill out this field
Please enter a valid email address.