Reducing fraudulent accounts by over 90% using Amazon Rekognition Face Liveness with Seeking.com

Since 2006, Seeking has grown into a worldwide community of millions of active members—sparking millions of relationships that are rooted in shared lifestyles and values. As online dating evolved, Seeking remained committed to offering its users a trustworthy platform with advanced security features.

Recognizing the sophistication of today’s digital landscape, the company wanted to proactively enhance its security measures beyond standard identity checks and IP address verification. “Our members expect a premium experience, and that begins with absolute trust,” says Hoony Youn, chief technology officer at Seeking. “Keeping pace with security standards isn’t good enough. We want to set the standards.”

Daniel Hartwig, director of product at Seeking, attended a third-party conference to learn how fraudsters tricked existing anti-fraud systems. Equipped with this knowledge, the company developed a multilayered approach to decreasing fraud. As a longtime user of AWS, Seeking decided to implement Amazon Rekognition Face Liveness, which can detect real users and deter bad actors who use spoofs in seconds during facial verification. Now, Seeking can spot face masks, printed photos, and deepfakes. “AWS lists all the ways the service can mitigate spoofing, so it was a simple choice,” says Hartwig.

Seeking rolled out the new approach after 2 months of development. Now, the Amazon Rekognition Face Liveness check is a required step to access the app, reducing the number of fraudulent accounts by over 90 percent.

Seeking also compares new user verification photos with previously banned accounts to detect banned users who try to sign up with different credentials. “It’s a game changer,” says Youn. “When we stop them once, we can recognize them when they try again.” Additionally, the company uses Amazon Bedrock, an easy way to build and scale generative AI applications with foundation models, to identify suspicious behavior patterns after the initial screening.

User feedback was positive, and the service had virtually no drop-off in use after adding the face-screening step. “Our job is to help people find love, and they should feel that love in the way they use Seeking,” says Brandon Wade, founder and co-CEO of Seeking. “We’re using AI and AWS to verify that people are on our platform for the right reasons.”

Seeking continues working to stay ahead of bad actors. “The last thing people should worry about is whether they’re communicating with someone real,” says Dana Rosewall, co-CEO of Seeking. “So, we’re trying to tackle fraud not only for our site but also for the whole industry.”