Metaverse privacy may not be possible without next-generation security measures, the findings suggest Researchconducted by scientists from the University of California at Berkeley.
The most surprising thing about the study was the fact that only very little data is required to identify a user in virtual space, so that real anonymity in the metaverse can be completely ruled out. Modern virtual reality headsets are equipped with numerous cameras, microphones and sensors that collect various data: user’s facial features, voice timbre, eye movements, as well as the surrounding objects in the home or office. In the future, it’s even possible to install EEG sensors that read indicators of brain activity through the scalp. And even if you remove this entire phrase, it is still not possible to ensure anonymity – to identify a person, it is enough to analyze simple data about his movement.
By “simple data” we mean the movements of three basic points tracked by virtual reality headsets: one on the user’s head and two on the user’s hands – this is the minimum data set required to work in a virtual environment is needed. Berkeley scientists analyzed 2.5 million anonymous recordings from more than 50,000 users of the Beat Saber VR app and found that they needed motion data recorded in just 100 seconds to identify a person with 94% accuracy. At the same time, half of the users could be identified using a 2-second piece of data. The study required the work of artificial intelligence algorithms, but it must be borne in mind that the data was extremely scarce – 2 seconds of movement from just three points.
This poses a serious threat to user privacy – anonymity in the Metaverse is therefore impossible. Aside from identifying an individual, this meager data is enough to determine specific information about an individual: height, gender, and dominant hand. And in combination with data from other sensors, the identification can become even more precise. When a person connects to the metaverse and visits a virtual furniture store, they make rather simple movements: grab virtual items from virtual shelves or step back a few steps to see what they look like. Berkeley scientists point out that all of these daily movements are unique to each person, just like their fingerprints.
Protecting user privacy in the Metaverse will be much more difficult. As one of the options, the researchers propose transmitting data from sensors to intermediate resources that falsify it. But that means information noise, reduced accuracy in VR headsets, and loss of enjoyment of a game that requires physical skill. There is also an alternative option – the introduction of industry regulations that prohibit the metaverse platforms from storing and analyzing human movement data. Such regulations are intended to protect the public, but enforcing such requirements will not be easy to enforce – industry will definitely resist. However, if consumers don’t feel safe in the metaverse, it will be difficult to make virtual spaces an important part of a person’s daily life.