Security researchers at the University of Toronto have developed a new attack that exploits the super-fast memory in a GPU to gain elevated privileges in Windows. The researchers used this method to leverage targeted data changes into read-write access on the GPU, which can lead to elevated permissions system-wide. The attack can also bypass input-output memory management. While this is a functional attack, it is only being used in a research lab. Nvidia may update its security guidance to enable Error Correcting Code memory features on industrial GPUs like the RTX A6000.