"Humans are super individual." We have different finger lengths, different joints, different motor skills. "Even if two people press the same key with the same finger, it takes one person longer than the other," explains Nils Vossebein, co-founder of DeepSign.
The startup takes advantage of this uniqueness. Using machine learning, people can identify themselves solely through computer use. "We connect the computer with the user," says Nils. It takes eight to ten interactions, such as a keystroke, a click, a mouse movement, to recognize a person. What emerges is a value that measures how likely it is that a legitimate user is sitting in front of the screen. Nils: "Valid users usually have a value of 90 to 95 percent. Other users are significantly below that. In the pilot phase, we defined anything below 80 percent as an attacker." If the computer then notices a stranger in front of the computer, it can automatically lock itself. "It's indeed very hard to imitate individual usage, especially when typing," Nils explains, adding, "Keyboard usage is recognized as a biometric feature."
One further step for DeepSign is feeding the system with information to the extent that a user is recognized even while riding a train, or when using different keyboards and mice. The Startup is also looking at the use of touchpads.
"The more often different devices and work locations have been used, the sharper the profile becomes. The system learns to recognize a person. You have to train it," says Nils.
Just recently, DeepSign found a pilot customer. Since March 2022, the technology has been in use at a company in the IT sector. 20 people will be testing the system for several months in order to find errors and provide indications for a simplified process.
So far, the technology is designed to replace a second factor in multi-factor authentication. "A lot of companies we've talked to have abolished the second factor because it was inconvenient," Nils tells us. So it's much more practical to just log in and be on the safe side. The computer knows you're in the right place.
