Security researchers have identified new attack techniques that exploit vulnerabilities in graphics processing unit memory to compromise computer systems running Nvidia GPUs.

The newly discovered attacks, designated as GDDRHammer and GeForge, target the memory systems of graphics cards to gain unauthorized access to central processing units. These techniques represent an evolution of memory-based attack methods that manipulate hardware components to bypass security measures.

Both attack variants work by deliberately stressing GPU memory in specific patterns that can cause system-level compromises. The techniques demonstrate how graphics hardware, traditionally considered separate from core system security, can become an attack vector for gaining complete machine control.

The discovery highlights growing concerns about hardware-level security vulnerabilities as computing systems become increasingly complex and interconnected. Graphics cards, which handle intensive computational tasks and have direct memory access, present new potential pathways for security exploits.

The research adds to ongoing discussions about the security implications of specialized computing hardware and the need for comprehensive security approaches that account for all system components, not just traditional CPU and operating system protections.