University of Toronto Researchers Demonstrate GPU-Based Rowhammer Attack
Researchers at the University of Toronto have demonstrated a new GPU-based Rowhammer attack, named GPUBreach, capable of escalating privileges to achieve full system compromise. This technique exploits memory corruption on modern graphics hardware to gain root-level access across both GPU and CPU environments. The attack focuses on corrupting GPU page tables using Rowhammer-induced bit flips in GDDR6 memory, allowing arbitrary read and write access to GPU memory. By targeting vulnerabilities in the NVIDIA driver, the attack extends to CPU memory, resulting in full system control. The research challenges existing assumptions about GPU security, highlighting the need for reassessment of current defensive measures. 
