Revolutionary Cybersecurity Threat: GPU Side-Channel Vulnerability Leaking Sensitive Data Unearthed

Revolutionary Cybersecurity Threat: GPU Side-Channel Vulnerability Leaking Sensitive Data Unearthed

unrecognizable-man-showing-weapon

Scientists from renowned academic institutions, including the University of Texas at Austin, Carnegie Mellon University, University of Washington, and the University of Illinois Urbana-Champaign, have disclosed a groundbreaking vulnerability in contemporary graphics processing units (GPUs). This recently discovered GPU side-channel attack, named GPU.zip, poses a serious risk to all major modern GPUs, revealing them to be vulnerable to alarming information leakage.

The newfound vulnerability pivots on exploiting graphical data compression—a functionality that enhances memory bandwidth conservation and subsequently amplifies the performance of integrated GPUs (iGPUs). However, this compression instigates data-dependent DRAM traffic and cache occupancy that can be measured via a side-channel, paving the pathway for potential cyber assaults.

Transitioning from the theoretical underpinnings to the practical implications of GPU.zip, the vulnerability permits a malevolent agent to exploit the iGPU-based compression channel to execute cross-origin pixel stealing attacks within the browser, capitalizing on SVG filters specifically. Pulling this off requires the cyber-attacker to generate either highly redundant or non-redundant patterns contingent on a solitary secret pixel, which consequently unduly influences the lossless compression output.

If successfully exploited, this vulnerability can facilitate a harmful web page to conjecture the values of individual pixels from another web page embedded in an iframe element in the latest version of Google Chrome. Disturbingly, this occurs while bypassing the same-origin policy (SOP). Both Google Chrome and Microsoft Edge are particularly susceptible to this innovative assault, but Mozilla Firefox and Apple Safari emerge as relatively resistant to the potential threat.

Proof-of-concept (PoC) implementation has demonstrated that this class of attack can deceive a user into visiting a rogue website, thus revealing information pertaining to the logged-in user’s Wikipedia username. Websites that incorporate protective measures, barring cross-origin websites from embedding them via X-Frame-Options and Content Security Policy (CSP) rules, possess immunity to this form of pixel-stealing assault. Still, the damage could be substantial without such safeguards in place.

Affected GPU produces encompass noteworthy names like AMD, Apple, Arm, Intel, Nvidia, and Qualcomm. The discovery of the GPU.zip vulnerability is documented in turn to another related side-channel attack—Hot Pixels—which aims particularly at Chrome and Safari browsers, enabling browser-based pixel stealing and history sniffing attacks.

In conclusion, the escalating sophistication of cyber-attacks, as evident from the GPU.zip case, underscores the critical importance of continual vigilance and responsive action within the cybersecurity domain. As cyber threats evolve, so must defense measures; thus, developers are urged to adequately fortify their websites and ensure they are impervious to such invasive attacks.


If you enjoyed this article, please check out our other articles on CyberNow

September 27, 2023
Scientists from renowned academic institutions have disclosed a groundbreaking vulnerability in contemporary graphics processing units (GPUs). This recently discovered GPU side-channel attack, named GPU.zip, poses a serious risk to all major modern GPUs, revealing them to be vulnerable to alarming information leakage.