Today, Amazon Web services (AWS) has launched a new tool, Amazon fraud detector, to help users identify fraudulent online activities such as online payment and identity fraud. The tool uses machine learning and Amazon’s more than 20 years of fraud detection expertise to detect fraud in milliseconds.
Amazon fraud detector is a fully hosted service that can easily identify potential fraudulent online activities, such as online payment fraud and the creation of false accounts.
billions of dollars of assets are defrauded by online fraud every year. Companies that do business online are particularly vulnerable to criminals. These criminals often use different attacks, such as creating fake accounts or paying with stolen credit cards. Companies often use fraud detection applications to detect fraudsters and stop them before they have a significant business impact on their business. However, these applications often rely on business rules, which can’t keep up with the speed at which fraudsters change their behavior and methods. Recent fraud detection applications have tried to use machine learning. However, they often use a one size fits all approach based on common data sets and fraud, which are not directly related to your business and therefore limited in accuracy.
fraud detector overcomes these difficulties by using your own data, machine learning (ML) and Amazon’s more than 20 years of fraud detection expertise. It can automatically identify potential online fraud, so you can detect more fraud faster. Even if you don’t have previous ml experience, you can create a fraud detection model with just a few clicks, because fraud detector handles all the ML heavy tasks for you.
accurately distinguish legal and high-risk customer account registrations so that you can selectively introduce other steps or checks based on risk. For example, you can set up the customer account registration workflow to provide additional e-mail and phone verification steps only for account registration requirements that exhibit high-risk characteristics.
potential fraudsters can be found even in customer groups with no transaction history. Customers who trade regularly usually use a registered account. As a result, you can get transaction history, which makes it easier to detect potential fraud. On the other hand, because guest checkout does not leave historical account usage or user behavior data, it is more difficult to detect fraud. With Amazon fraud detector, you can assess the potential fraud risk by simply sending an e-mail and IP address from a guest bill, so you can decide whether to accept an order, review an order, or collect more customer details.
we found that it was more likely to abuse the “try before buy” service account, such as fashion services that sent clothing and accessories to you for experience before paying. Through Amazon fraud detector, online businesses can assess the risk of customers violating the terms of service, and set certain limits on the value of goods or services provided to prevent goods from being stolen or returned due to violation of the terms of service.
reduce online payment fraud by tagging suspicious online payment transactions before processing payments and completing orders. With Amazon fraud detector, you can set up the checkout process to evaluate new orders and mark suspicious orders before processing payments, reducing credit card refunds.
Amazon fraud detector provides templates that make it easy for you to create ml models, so you don’t need to write any code to detect potential fraud. You simply upload historical online event data, such as transactions or account registrations, and select the model template that matches your use case. On this basis, Amazon fraud detector can automatically train, test and deploy custom fraud detection models suitable for your business.
when fraudsters create accounts, Amazon fraud detector can immediately detect fraudsters by predicting the risks in the information they provide, so that you can mark suspicious activities before they cause actual losses. This is because Amazon fraud detector uses advanced machine learning technology, which can even be used to provide limited data when creating accounts. Amazon fraud detector’s machine learning model can detect up to 80% of potential criminals than traditional methods.
the pre built machine learning model template in Amazon fraud detector was developed based on more than 20 years of experience in preventing criminals from trying to cheat AWS and. For example, Amazon fraud detector uses a model similar to that in the AWS account registration process to create different account validation steps for low-risk and high-risk registrations.
Amazon fraud detector automates the complex tasks required to build, train, tune, and deploy fraud detection models, enabling the anti fraud team to move faster. After creating the model, they are able to create, view, and update rules to enable model prediction based operations without relying on the help of others.