Your graphic card can reveal your location and browsing activities within 8 seconds through a newfound hack

Researchers have come up with a novel method for digital fingerprinting that targets the GPU in a system. The method is able to track the location of the targeted users as well as their online activities better than any existing tracing technology.

There has been a gigantic increase in digital fingerprints of an average internet user with rapid digitisation. A new study now adds a new and a rather unconventional method to this list. It states that users can be tracked down through the graphics cards they use on their systems.

The research comes from an international team of researchers and has now been published on the online open-access archive Arxiv.org. It talks of a new technique called “GPU fingerprinting” that can help locate a target by mapping the GPU performance of their system.

The technology that enables this tracking has been named DrawnApart by the researchers. It works by mapping the tiny differences between each GPU unit. Since all graphic cards have these minute differences, the technique is able to distinctly mark a unit by analysing these and hence track down a target system and its user.

In their paper, researchers state that the technique makes it possible to track down individual users and even map their online activity. As spotted by Bleeping Computer, the paper also showcases a real-life demonstration of this GPU fingerprinting technique. Gaining highly accurate results from the technique, the researchers actually found it to be an upgrade over the existing fingerprinting techniques.

In fact, DrawnApart worked so well that it enabled the fingerprinting of a graphics card within eight seconds. This was noted to be a 67 per cent improvement over the current fingerprinting methods other than DrawnApart. The results originated from a sample size of over 2,500 unique devices and 371,000 fingerprints.

As mentioned, DrawnApart relies on the minute differences between hardware components, even if they are of the same make. It uses a cross-platform JavaScript-based application programming interface (API), which is able to render graphics on compatible web browsers.

These APIs target the GPU’s shaders with a sequence of graphic operations specially designed for the task. The responses from these Execution Units (EUs) are then timed for each rendering task, which ultimately results in a precise fingerprint trace. The paper highlights that the technique can even differentiate between units of the same make and model.