Back Scworld

Total takeover of Nvidia GPU-based devices possible with novel Rowhammer attacks

Scworld • April 6, 2026

Machines running high-performance Nvidia GPU cards, which are prevalent in cloud environments, could be completely hijacked through three new Rowhammer intrusions, Ars Technica reports.

Both the GDDRHammer and GeForge attacks discovered by two different groups of researchers could target Nvidia RTX 6000 chips using the Ampere architecture to trigger GDDR bitflips that result in total CPU memory takeovers, provided that IOMMU memory management is deactivated. Despite Rowhammer protections on the CPU, attackers could use GDDRHammer to Rowhammer the CPU and later compromise the CPU, according to researcher Andrew Kwong. New hammering patterns and memory massaging techniques have also been leveraged by GeForge for GPU page table mapping corruption in GDDR6 memory.

On the other hand, GPUBreach, which was uncovered by another set of researchers, harnesses GPU driver-based memory safety vulnerabilities to elevate privileges to root even with IOMMU enabled. While no Rowhammer attack has been observed in the wild so far, such findings emphasize the potential security risks faced by GPUs.

The full story

This article is one source in a clustered incident — the cluster page carries the summary, timeline and every other outlet covering it.

GPUBreach: New GPU Rowhammer Attack Enables Full System Compromise Researchers from the University of Toronto have unveiled a new attack named GPUBreach that exploits Rowhammer vulnerabilities in Nvidia's GDDR6 memory. This attack allows privilege escalation from an unprivileged CUDA kernel to root access on the host system without needing to disable the Input-Output Memory Management Unit (IOMMU). By corrupting GPU page tables, attackers can gain arbitrary read/write access to GPU memory and subsequently exploit memory-safety bugs in the Nvidia driver to achieve full system compromise. The attack can manipulate sensitive data, including cryptographic keys and machine learning model weights, significantly impacting AI workloads. The findings will be presented at the IEEE Symposium on Security & Privacy on April 13, 2026. Nvidia has acknowledged the issue and may update its security advisory, while Google awarded the researchers a $600 bug bounty for their discovery. Despite the severity, the current risk level is considered manageable by some industry players. 11 sources 67

Extracted Entities

Total takeover of Nvidia GPU-based devices possible with novel Rowhammer attacks

The full story

Extracted Entities

Attack Types (1)

Platforms (5)

Tools (3)

Vulnerabilities (1)