Using off-the-shelf gaming technology that tracks brain activity, a team of scientists has shown that it's possible to steal passwords and other personal information.
Researchers from the University of Oxford, University of Geneva and the University of California at Berkeley demonstrated the possibility of brain hacking using software built to work with Emotiv Systems' $299 EPOC neuro-headset.
Developers build software today that responds to signals emitted over Bluetooth from EPOC and other so-called brain computer interfaces (BCI), such as MindWave from NeuroSky. Of course, if software developers can build apps for such devices, so can criminals.
"The security risks involved in using consumer-grade BCI devices have never been studied and the impact of malicious software with access to the device is unexplored," the researchers said in a paper presented in July at the USENIX computer conference. "We take a rst step in studying the security implications of such devices and demonstrate that this upcoming technology could be turned against users to reveal their private and secret information."
The researchers found that the software they built to read signals from EPOC significantly improved the chances of guessing personal identification numbers (PINs), the general area participants in the experiment lived, people they knew, their month of birth, and the name of their bank.
The Emotiv device, used in gaming and as a hands-free keyboard, uses sensors to record electrical activity along the scalp. Voltage in the brain spikes when people see something they recognize, so tracking the fluctuation makes it possible to gather information about people by showing them series of images.
The researchers conducted their experiments on 28 computer science students. In the PIN experiment, the subjects chose a four-digit number and then watched as the numbers zero to nine were flashed on a computer screen 10 times for each digit. While the images flashed before the subjects, the researchers tracked brain activity through signals from the EPOC neuro-headset.
The same form of repetitive showing of images was used in the other experiments, such as a series of bankcards to determine a subject's bank or images of people to find the one they knew.
In general, the researchers' chance of guessing correctly increased to between 20% and 30%, up from 10% without the brain tracking. The exception was in figuring out people's month of birth. The rate of guessing correctly increased to as much as 60%.
Nevertheless, the overall reliability was not high enough for an attack targeted at a few individuals. "The attack works, but not in a reliable way," Mario Frank, a UC Berkeley researcher in the study, said on Friday. "With the equipment that we used, it's not possible to be sure that you found the true answer."
A criminal would have to build malware that could be distributed to as many people as possible. Such a tactic is used in distributing malware via email, knowing that only a small fraction of recipients will open the attachments. However, that small fraction is enough to create botnets of hundreds of thousands of computers.
With BCI devices, the user base today is too small to launch large-scale attacks. Also, users buy software directly from manufacturers, so it would be difficult for criminals to distribute malware.
However, a security risk could arise in the future, if brain-tracking devices become standard for interacting with computers and online stores are created to sell hundreds of thousands of applications, much like people buy apps for Android smartphones today.
To minimize risk, device manufacturers should start building security mechanisms today, such as limiting the information software can access from the headset to only the data needed to run the app, experts say.
"One thing that could be improved, for instance, is that the device itself does some pre-processing and only outputs the data that is required for the application," Frank said.
Such precautions should be taken today to prevent unnecessary risks in the future, he said.