IT House News on February 1, in a research paper titled "DRAWNAPART: Device Recognition Technology Based on Remote GPU Fingerprint Recognition", researchers proposed a method for identifying users based on GPU-generated browser "fingerprints".
Currently, websites can track users through cookies, browser user agents, web IPs, mouse movements, and other technologies to provide more accurate advertising. However, researchers have confirmed that there is a new high-precision way to identify users: GPU fingerprints.
In large-scale experimental tests involving 2500 devices, tracking was 67% faster than other known methods. Current approaches that rely on WebGL 2.0 APUs require at least 8 seconds to recognize the GPU, but new Web APIs have reduced this time to 150 milliseconds and increased accuracy to 98%.
According to reports, this method relies on hardware recognition, especially GPUs. The researchers took advantage of the possibility of measuring vectors containing various calculations that could be used to validate users. Researchers have even confirmed that removing and replacing certain components does not affect the "classifier" used to track users.
However, this method of identification requires a WebGL 2.0 API, and only a few sites such as Google Maps or IKEA are still using WebGL 2.0. Only 1% of Alexa's top 10K sites require the WebGL standard, which means that browsers can disable this API by default.
IT House understands that Khronos, the nonprofit organization responsible for WebGL, is already working on possible mitigations to stop this identification method.