Breaking
OpenAI releases GPT-5 — shatters every benchmark, approaches human-level reasoning on MMLU at 92.4% ◆ NVIDIA Blackwell GPUs sold out through 2026 as AI data centre demand hits record highs ◆ US Government issues landmark AI Executive Order — new compliance rules for foundation model labs ◆ Google Gemini Ultra 2.0 launches for enterprise — native integration across Workspace and Cloud ◆ Anthropic raises $4B Series E at $60B valuation, doubles safety research headcount ◆ EU AI Act enforcement begins — Apple, Google, and OpenAI face first wave of compliance deadlines ◆ AI startups raise $42B in Q1 2025 — a new global record; healthcare and robotics lead verticals ◆ Meta releases Llama 4 open-source: matches GPT-4 performance, free for commercial use      OpenAI releases GPT-5 — shatters every benchmark, approaches human-level reasoning on MMLU at 92.4% ◆ NVIDIA Blackwell GPUs sold out through 2026 as AI data centre demand hits record highs ◆ US Government issues landmark AI Executive Order — new compliance rules for foundation model labs ◆ Google Gemini Ultra 2.0 launches for enterprise — native integration across Workspace and Cloud ◆ Anthropic raises $4B Series E at $60B valuation, doubles safety research headcount ◆ EU AI Act enforcement begins — Apple, Google, and OpenAI face first wave of compliance deadlines ◆ AI startups raise $42B in Q1 2025 — a new global record; healthcare and robotics lead verticals ◆ Meta releases Llama 4 open-source: matches GPT-4 performance, free for commercial use
Back to News
CybersecurityBearish SignalHigh Impact

Nvidia GPUs vulnerable to devastating hijack attacks

Share: X LinkedIn WhatsApp

Security researchers have discovered new Rowhammer attacks that can give attackers complete control of machines running Nvidia GPUs, highlighting the ongoing vulnerabilities in computer hardware. The Blacksmith and HammerDL attacks exploit weaknesses in DRAM and can bypass existing mitigations, demonstrating the need for continued innovation in mitigation strategies.

Nvidia GPUs vulnerable to devastating hijack attacks
AE
AnalyticsGlobe Editorial
AI & Technology Desk
24 April 20266 min read9 views

Security researchers have discovered new Rowhammer attacks that can give attackers complete control of machines running Nvidia GPUs, highlighting the ongoing vulnerabilities in computer hardware.

Background

Rowhammer attacks exploit a weakness in dynamic random-access memory (DRAM) that allows attackers to flip bits in adjacent rows by repeatedly accessing, or "hammering," the same row. This can lead to privilege escalation, allowing attackers to gain control of a system.

Vulnerability Details

The new Rowhammer attacks, dubbed "Blacksmith" and "HammerDL," specifically target Nvidia GPUs and can bypass existing mitigations. Blacksmith uses a novel approach to induce bit flips, while HammerDL leverages machine learning to optimize the attack.

Key Facts

  • Blacksmith and HammerDL attacks can compromise the security of machines running Nvidia GPUs.
  • These attacks exploit weaknesses in DRAM and can bypass existing mitigations.
  • The vulnerabilities can be used to gain complete control of a system.

Industry Analysis

The discovery of these new Rowhammer attacks highlights the ongoing need for vigilance in the field of computer security. As hardware vulnerabilities continue to be exploited, manufacturers must prioritize the development of secure hardware and software mitigations.

While the Blacksmith and HammerDL attacks are significant, they also demonstrate the importance of ongoing research into hardware security and the need for collaboration between industry and academia to address these vulnerabilities.

Expert Perspective

According to security experts, the Blacksmith and HammerDL attacks demonstrate the evolving nature of Rowhammer exploits and the need for continued innovation in mitigation strategies. Machine learning-based approaches may hold promise for detecting and preventing these types of attacks.

Conclusion

In conclusion, the new Rowhammer attacks targeting Nvidia GPUs underscore the importance of prioritizing computer security and the need for ongoing research into hardware vulnerabilities. As the tech industry continues to evolve, it is essential to stay ahead of emerging threats and develop effective mitigations to protect against them.

Tags:Nvidia GPUsRowhammer attackscomputer securityhardware vulnerabilitiesmachine learning
Disclaimer

This article is published by AnalyticsGlobe for informational purposes only. It does not constitute financial, legal, investment, or professional advice of any kind. Always conduct your own research and consult qualified professionals before making any decisions.

AE

AnalyticsGlobe Editorial

AI & Technology Desk

Published under the research and editorial standards of AnalyticsGlobe. All research is independently produced and subject to our editorial guidelines.