Scams suck.
It’s a universal truth that we all have some degree of experience with. At best, encountering a scammer is merely an annoyance or a waste of time. At worst, victims are left humiliated and penniless, having been manipulated into giving up their hard-earned savings.
Meet Serket
Serket is an AI dispatcher that pre-screens phone calls and uses voice identification to prevent scammers from reaching a point of contact with their intended targets.
Design Process
Spam: those obnoxious robocalls trying to market products to you (your car’s warranty is almost certainly fine)
Scam: a targeted scheme which uses emotion, urgency, and personal data to extract money from the victim
Phone scammers specifically target seniors, taking over 3 billion dollars from them each year
Federal Bureau of Investigation
After speaking with scam experts in academia and the Department of Justice, as well as victims themselves, we developed the following persona to better understand our users. Cathy was the victim of a grandparent scam, which convinces someone to send emergency funds to a supposed grandchild who is in trouble.
We’ll see Cathy again later as part of the Serket solution, where the very things that made her a prime target are used to turn the tables on potential scammers.
Currently, people targeted by scammers have difficulty determining the validity of the call, even when assisted by technology. Once the scam takes place, it’s also a challenge to report and prosecute the crime.
We designed Serket to protect stakeholders before they ever connect to a potentially dangerous caller.
When receiving a call, Serket rapidly runs a series of checks to ensure optimal friction depending on the caller’s motivation.
Expected calls from a delivery service or trusted contact, for example, are patched through immediately, posing no interruption to users on either end of the line. Unknown callers are asked to verify their identity using voice biometrics, which are validated against our database of scammer voices, and can also be further vetted using custom questions from the call recipient.
Remember our user persona Cathy Bates?
“Bait” identities like hers are used to entice scammers into calling fake numbers, where their voice data is captured and added to the Serket database. This allows us to train the voice biometric models and identify scammers later on, while simultaneously protecting real potential victims.
Like many data-driven solutions, Serket is designed to continuously improve over time. As more scammer data is collected, we will be able to block scammers with even more speed and accuracy. Through partnerships and network analysis, we also hope to help prior victims prosecute their scammers and find the justice they deserve.
