So we have some good and bad news for you.
The good news is that most major companies are doing away with CAPTCHA — you know that annoying letter puzzle which apparently proves you aren’t a robot.
Bad news… they are getting rid of it because China already has a way to easily get past such robot tests.
To be more specific, a CAPTCHA is a program or system used in computing to determine whether a user is a human or a robot. The most common type is text-based, where users have to identify jumbled letters and numbers to prove they are human. There’s also a newer version where users are asked to identify common items in a series of images.
The goal is to prevent bots and spammers from accessing information.
As reported in the blog Naked Security, a team of researchers from the US, UK, and China used a generative adversarial network (GAN) to solve text-based CAPTCHAs.
A GAN is a relatively new type of neural network that learns from examples. After looking through just 500 CAPTCHA examples, the network was able to solve a real-world CAPTCHA in less than one second.
The team tested their program against CAPTCHAs used by 11 big Internet companies with various success rates:
- eBay (87%)
- Wikipedia (78%)
- Microsoft (70%)
- Google (3%)
“We show for the first time that an adversary can quickly launch an attack on a new text-based CAPTCHA scheme with very low effort,” writes Lancaster University’s Dr. Zheng Wang. “This is scary because it means that this first security defense of many websites is no longer reliable.”
Based on the team’s success, we can expect websites to stop using text-based CAPTCHAs.
In case you’re wondering, CAPTCHA stands for “Completely Automated Procedures for Telling Computers and Humans Apart.”