# hardware
3 articlestagged with “hardware”
GPU Memory Side-Channel Attacks
Side-channel attacks exploiting GPU memory allocation, timing, and electromagnetic emanation to extract sensitive data from AI workloads.
infrastructuregpuside-channelprivacyhardware
Hardware Security for ML Accelerators
Hardware-level security considerations for ML accelerators including side-channel attacks, firmware vulnerabilities, and memory protection.
infrastructurehardwareacceleratorsTPU
Lab: GPU Side-Channel Attacks
Demonstrate information leakage through GPU memory residuals and timing side channels, showing how shared GPU infrastructure can expose sensitive model data.
labgpuside-channelhardwareinformation-leakage