Skip to main content

Amazon pilots face recognition for prospective third-party sellers

man on computer
U.S. Amazon sellers can now use facial recognition for identity verification.

Amazon is testing a voluntary program that uses biometric data to verify the identity of sellers applying to utilize its online marketplace.

As of Nov. 15, 2022, the e-tail giant is starting to test a new, optional solution for prospective sellers that uses forgery detection, facial recognition, and liveness detection technology to verify the authenticity of government-issued identity documents and whether they match the individual applying to sell in the Amazon online store.

Identity verification is a key step potential sellers must take before setting up shop on the Amazon platform. Amazon says this test will determine whether it can achieve the same robustness of identity verification as its current processes, which use a combination of proprietary technology and expert human reviews, while also providing a better experience for sellers.

Facial recognition – how it works
If a prospective Amazon seller opts to use facial recognition to verify their identity, they will first provide access to their device’s camera, and will be shown a series of steps in which images of their face and government-issued identity document are captured.

Next, the images will be analyzed using machine learning models to create unique facial measurements called facial biometric data. This data is processed in real time to confirm that the prospective seller’s face matches the face on their government-issued identification document.

Once that process is complete, the prospective seller will see a pending verification page. During the test period, Amazon will be manually reviewing the system’s decisions and verifying seller identities, and notifying prospective sellers of the decision within a few days. In the future, Amazon says many of these decisions will be automated and communicated instantly.

Security steps
To help ensure privacy and security for applicants, Amazon will only use facial biometric data captured through this test in real time to verify a seller’s identity by determining whether they are a live person, and whether their face matches the picture on their government-issued identification document.

The facial biometric data captured through this test will not be used for any purpose other than to verify a prospective seller’s identity, Amazon said. Images captured through this test will be encrypted both in transit and at rest.

If prospective sellers prefer, they can complete identity verification through the current process of an in-person video call. Amazon says the pilot complies with all applicable laws, such as the Biometric Information Privacy Act (BIPA) and California Consumer Privacy Act (CCPA).

Amazon is starting this test with prospective sellers in its U.S. store, and will gather feedback from sellers to help improve the experience over time and determine if it can provide a better experience for seller verification.

The rise of Amazon sellers
Third-party sellers are becoming an increasingly important part of Amazon’s business. According to the company, during its 2022 Prime Day event, which broke records with close to $12 billion in total sales during the July 12-13 period, was also the biggest-ever Prime Day event for Amazon's third-party selling partners. Amazon said sales growth by its selling partners outpaced growth in its own retail business, with Prime Day shoppers spending over $3 billion on more than 100 million small business items.

We take pride in creating a trustworthy shopping experience that allows customers to shop with confidence,” Anirudh Chandrakant, Director of Seller Registration Services at Amazon, said in a corporate blog post. “As we continue to iterate upon and improve the selling experience, some small businesses have told us they would like an even more efficient way to complete our identity verification process, one of the key steps before someone can begin selling in our store.”

This ad will auto-close in 10 seconds