Rowhammer attacks have been extensively studied on CPUs, where they have enabled powerful exploits, including privilege escalation. In contrast, Rowhammer on NVIDIA GPUs, despite recently demonstrated by GPUHammer attacks, has largely been viewed as low impact, limited to inducing random bit flips that merely degrade machine learning accuracy. In this Briefing, we will overturn that assumption and show that GPU Rowhammer can be weaponized into a full-system compromise.
We will demonstrate GPUBreach, the first targeted Rowhammer attacks on NVIDIA GPUs that enable privilege escalation and cross-process memory access. By reverse engineering GPU page table allocation behavior, we identify when and where page tables are allocated, allowing an unprivileged CUDA kernel to induce deterministic bit flips in victim GPU page table entries. This enables corruption of GPU page tables and yields arbitrary read and write access to GPU memory across isolation boundaries. We use these primitives to demonstrate concrete attacks, including the extraction of cryptographic keys from GPU-accelerated libraries like cuPQC, the exfiltration of proprietary machine learning model weights, and the stealthy modification of GPU kernel code (e.g., in cuBLAS) to universally degrade the accuracy of any ML model.
Most critically, we show that GPU compromise does not stay contained. We bridge the GPU-CPU boundary to achieve full system takeover despite IOMMU protections that are designed to protect the host against a malicious device. Leveraging GPU-side arbitrary writes to inject malicious DMA operations to the CPU memory, we exploit a previously unknown memory safety vulnerability in the NVIDIA GPU driver stack, and ultimately gain arbitrary write access in the Linux kernel, resulting in root shell access.
These results fundamentally challenge the security assumptions of modern heterogeneous systems and demonstrate that GPUs are not just an accelerator, but powerful attack surfaces that break the security of the entire system.
Takeaway
1. **Turning GPU Rowhammer into a targeted exploitation primitive capable of privilege escalation:** Attendees will learn how seemingly random GPU Rowhammer bit flips can be engineered into precise, attacker-controlled faults. We will walk through how to reverse engineer GPU page table structures and allocation behavior on NVIDIA GPUs, identify when and where page tables are placed in memory, and use memory "massaging" techniques to position them in vulnerable regions. These insights enable deterministic corruption of page table entries, yielding arbitrary read/write access to GPU memory and breaking isolation across processes and containers.
2. **Exploiting New NVIDIA GPU Driver Memory Safety Vulnerabilities.** Attendees will gain a practical understanding of how GPU compromise can be extended to the host. We will show how GPU-controlled DMA can be used to tamper with privileged data structures in the NVIDIA Driver on the host. Through a deep-dive of the CPU-GPU communication buffers in the open-source NVIDIA GPU Driver, we will show how trusted-but-unsanitized inputs from the device side lead to out-of-bounds writes within the drivers, and how this can be turned into powerful primitives such as arbitrary kernel writes and function pointer overwrites within the Linux kernel, that can lead to system-wide privilege escalation.
3. **Rethinking GPU Security Assumptions:** This talk encourages attendees to rethink traditional GPU trust models. For GPU cloud operators and users, this highlights the immediate need to adopt recommended Rowhammer mitigations such as enabling ECC on all GPUs. For Linux kernel and device driver developers, this encourages a need to rethink the trust-assumption in GPUs, harden the drivers against a new threat of "malicious" GPUs. We will discuss practical defensive strategies, ranging from hardware and firmware mitigations against Rowhammer-induced page-table corruption to software techniques for hardening drivers and systems against compromised or malicious GPU devices.
These insights help security practitioners understand the real attack surface of GPUs and evaluate risks in AI infrastructure, workstations, and shared GPU environments.
Chris S. Lin | PhD Student, University of Toronto
Yuqin Yan | PhD Student, University of Toronto
Guozhen Ding | Master Student, University of Toronto
Joyce Qu | Undergraduate Student, University of Toronto
Joseph Zhu | Machine Learning Engineer, Google
David Lie | Professor and Canada Research Chair, University of Toronto
Gururaj Saileshwar | Assistant Professor, University of Toronto
https://ift.tt/wzNdQav
source https://www.youtube.com/watch?v=G-pFmwvADbY
The Cyber Stream
Latest News for Cyber Security & Technology
Thursday, 19 November 2026
Wednesday, 18 November 2026
Black Hat USA 2026 | Can AI Do Novel Security Research? Meet the HTTP Terminator
We all know AI can find bugs. After a decade of research, I asked a harder question: can an autonomous system invent new attack techniques, and use them to hack live websites at scale? Building this sounded like a bad idea, so I did it.
It worked - I'll share an arsenal of new HTTP desync triggers, gadgets, and exploits that compromised banks, security solutions, and government infrastructure. Then I'll trace each discovery chain back through the HTTP Terminator, showing how to turn your personal expertise into an autonomous weapon - and the dark arts required to make it lethal.
I'll also share discoveries from beyond the autonomy horizon - some only reachable with a tight human/AI research loop, and others beyond AI's reach entirely. These include a powerful undisclosed recon technique, and anomalies that hint at new attack classes offering alternative paths to critical impact. I'll analyze the discovery process, sharing detailed experiments that probe the boundaries of what AI can and can't discover.
You'll leave with new exploits from desync triggers to undisclosed attack classes, and a blueprint for turning your instincts into an autonomous research cascade. And yes, I'll open-source the HTTP Terminator.
Takeaway
Under Embargo.
James Kettle | Director of Research, PortSwigger
https://ift.tt/f5hLWVD
source https://www.youtube.com/watch?v=bERmUEJhWFM
source https://www.youtube.com/watch?v=bERmUEJhWFM
Tuesday, 17 November 2026
Black Hat USA 2026 | Threat Modeling LLMs: The PHANTOM-B model
Security engineers shipping LLM products face a painful mismatch: the threat landscape is complex and evolving, the boss wants it deployed yesterday, and existing resources — MITRE ATLAS, NIST AI RMF, hundreds of academic papers — are built for thoroughness, not speed.
PHANTOM-B is a practical threat modeling mnemonic developed and validated with hyperscalers and globally significant financial institutions. Its eight threats are grounded in cognitive psychology research on chunking and expert scaffolding. They're scoped to what engineering teams can actually influence within a real deployment timeline.
Unlike OWASP LLM Top 10, PHANTOM-B is structured as a threat elicitation tool, not a vulnerability list — designed to answer "what can go wrong in this system" rather than "what goes wrong in LLMs generally." Attendees will leave with the mnemonic, a repeatable facilitation approach, and a framework for shifting defensive strategy as LLM-specific risks evolve.
Takeaway
• A practical threat elicitation tool (PHANTOM-B) that a team can apply to any LLM deployment in under an hour
• A structured approach to shifting defenses as LLM-specific risks evolve: what changes in your PSIRT process, and your product team's responsibilities
• Confidence that the overwhelming volume of AI security research can be navigated — and a map for where PHANTOM-B fits relative to OWASP, ATLAS, and your existing SDL
Adam Shostack | Founder, Shostack + Associates
https://ift.tt/agy0YN6
source https://www.youtube.com/watch?v=A8H06BYjEq4
source https://www.youtube.com/watch?v=A8H06BYjEq4
Black Hat USA 2026 | LANJack: Turning Ads into IoT Recon Tools
You visit a legitimate website. A trusted brand advertisement loads. Nothing looks suspicious. No phishing page, no exploit kit, no visible signs of an attack. Meanwhile, your browser is silently scanning your internal network, mapping your LAN and identifying connected devices.
We present LANJack, the first known large-scale DNS rebinding campaign delivered through mainstream programmatic advertising infrastructure. Distributed as a branded creative through a legitimate DSP, the attack required no clicks and executed silently within the publisher's context, turning the browser into an internal reconnaissance platform. What began as a routine redirect investigation uncovered a structured attack framework operating at internet scale.
LANJack weaponized DNS rebinding to enumerate local subnets, identify hosts, fingerprint IoT devices, and interact with exposed services using vendor-specific logic across devices from major vendors including Hikvision, Dahua, UniView, TP-Link, Linksys, and HP.
Beyond internal network access, the same execution model was extended to infer authentication state across major platforms, demonstrating how browser-based attacks can move from local reconnaissance to cross-origin account intelligence.
This research shows that purchasing ad inventory is sufficient to deploy large-scale internal network attacks, leveraging known techniques and gaps in browser and network security assumptions. It exposes a structural blind spot between ad security and network security, where programmatic advertising becomes infrastructure for active network-level attacks.
Takeaway
1. Understand how DNS rebinding can be operationalized at scale through programmatic advertising, including techniques for browser-based LAN scanning, cache priming, and device fingerprinting without user interaction.
2. Identify why existing browser and network defenses fail against this attack model and recognize the gaps in controls such as same-origin policy, CSP, and network isolation when facing browser-originated access to internal resources.
3. Apply a new threat model that treats ad-delivered code as a potential internal network attack vector and use it to reassess risk in areas such as ad security, enterprise browser policies, and IoT exposure.
Moriya Pedael | Security Researcher, GeoEdge
https://ift.tt/hxedi4r
source https://www.youtube.com/watch?v=5HpphRnlzxk
source https://www.youtube.com/watch?v=5HpphRnlzxk
Wednesday, 4 November 2026
Fromage-Gate: Old-School Hacking Tactics vs. French Supermarkets
Who You See:
Daniel Cuthbert
-LinkedIn: https://ift.tt/gust7OL
-X: https://x.com/dcuthbert
The Grugq
-LinkedIn: https://www.linkedin.com/in/grugq/
-X: https://x.com/thegrugq
Show Notes:
Grain of rice
-Shorthand in the community for sensational, unproven hardware backdoor claims.
Referenced Talks:
Remote Exploitation of an Unaltered Passenger Vehicle
-Charlie Miller and Chris Valasek
-Black Hat USA 2015
https://youtu.be/MAcHkASmXEc?si=3X3jnzCBIS1155i1
MIDImorphosis - Exploiting a Windows 11 Privilege Escalation Vulnerability with a Teensy USB Device
-Andy Davis
-Black Hat Europe 2026
-https://ift.tt/fHrJuNC
Phantom Bytes: A Complete Taxonomy of Linux Memory Attacks Against Modern EDR
-Yuan Zhang
-Black Hat Europe 2026
-https://ift.tt/NGikqdy
Defeating Forensic Analysis on Unix
-Thaddeus Grugq
-Phrack Magazine 2002
-https://ift.tt/7jOgGLt
Remote Exec
-Thaddeus Grugq
-Phrack Magazine 2004
-https://ift.tt/JTXnWtb
Vivisection of an Exploit Development Process: What To Do When It's Not Easy
-David Aitel
-Black Hat Europe 2023
-https://ift.tt/fHWyeam
Professional Source Code Auditing
-Mark Dowd, Halvar Flake, Nishad Herath, Neel Mehta, and Chris Spencer
-Black Hat USA 2002
-https://youtu.be/YceCSBLlelQ?si=C8qhb2M9GcTA-AGt
Why We are Not Building a Defendable Internet
-Halvar Flake
-Black Hat Asia 2017 Keynote
-https://youtu.be/PLJJY5UFtqY?si=nesqa__4D0KQJAzl
00:00 Intro
00:14 Fromage-Gate with The Grugq
01:28 Industry Lag Since 1990
04:42 Cheesecake Factory- Early Script Kiddie Era
07:03 Curiosity vs Monetization
12:36 Reflections on the Early Hacking Community
16:01 LLMs Lowering the Barrier, Misuse- Ty & Calif, Anthropic
18:27 Absorptive Capacity- What Can be Weaponized
21:44 Notable Black Hat EU Research
24:34 Historical Research Worth Revisiting
39:54 Closing and Call to Action
Spotify: https://ift.tt/BjOWrCw
Apple Podcasts: https://ift.tt/BrN5gYV
source https://www.youtube.com/watch?v=QH9Lh7wVZAg
source https://www.youtube.com/watch?v=QH9Lh7wVZAg
Tuesday, 3 November 2026
SecTor 2026 | Using Claude or Codex? We Can Hack Your App?
Today, most code is actually vibe-coded by the same top AI code models, which often reuse the same implementation patterns, including vulnerability patterns. We found that an attacker can generate and inspect their own applications, discover vulnerabilities that developers may have missed, and use those findings to attack other applications generated by the same model, which have a high chance of containing the same vulnerabilities. In this Briefing, we evaluate how high these chances are among today's top code models and present a new attack and threat model.
We introduce FSTab, a novel black-box attack that maps visible UI features such as login, upload, and password reset to recurring corresponding backend vulnerabilities. We evaluate how often these vulnerabilities transfer across applications, prompts, and domains using six widely used code LLMs. FSTab reaches up to 85.67% average attack success and 84.97% average vulnerability coverage, showing how model-level coding habits can expose vulnerabilities across thousands of AI-generated applications.
Amit LeVi | AI Researcher & Founder, ATLAS Lab
Raz Tel-Vered | AI and Security Research Team Lead, Zenity
https://ift.tt/8lmx5Nk
source https://www.youtube.com/watch?v=y_lw2P0393k
source https://www.youtube.com/watch?v=y_lw2P0393k
Monday, 2 November 2026
Black Hat Europe 2026 | Welcome Video
Bringing together the brightest minds in cybersecurity, Black Hat is where the global security community comes to discover groundbreaking research, explore emerging threats, and exchange ideas that shape the future of the industry.
As AI transforms the digital ecosystem and the threat landscape evolves at an unprecedented pace, collaboration and knowledge sharing have never been more important.
This welcome video highlights everything you need to know before arriving onsite, from programming and venue information to networking opportunities and ways to get the most out of your Black Hat Europe experience.
Whether you're attending for the first time or returning as part of our community, we're excited to welcome you to London this December.
#BlackHat #cybersecurity #BHEU
https://ift.tt/NY8fndS
source https://www.youtube.com/watch?v=1HU_OsgzBQ0
source https://www.youtube.com/watch?v=1HU_OsgzBQ0
Wednesday, 21 October 2026
Black Hat Stories | Island.io
Hear Chris Gaun reflect on his experience as a first-time exhibitor at Black Hat.
Ready to be part of the conversation? Become a Black Hat sponsor: https://ift.tt/OFzpBy2
source https://www.youtube.com/shorts/eO0FZm_xeSo
source https://www.youtube.com/shorts/eO0FZm_xeSo
Tuesday, 20 October 2026
Black Hat Europe 2026 | Hear From Our Sponsors
Black Hat Europe brings together cybersecurity innovators and leaders, creating opportunities to showcase solutions and build meaningful relationships.
Hear directly from some of the brands that sponsored Black Hat Europe 2025. The connections they made, the conversations that mattered, and why the show floor energy keeps them coming back.
Ready to join them in 2026? Explore sponsorship opportunities: https://ift.tt/xMFIAHL
source https://www.youtube.com/watch?v=Muk6elQ_qcs
source https://www.youtube.com/watch?v=Muk6elQ_qcs
Black Hat Stories | First Time Exhibitor
On this episode, Chris Gaun, VP of Product Marketing at Island.io, shares his experience exhibiting at Black Hat for the first time.
Ready to be part of the conversation? Become a Black Hat sponsor: https://ift.tt/fuTPJ1m
source https://www.youtube.com/shorts/NAPi2gWNZP4
source https://www.youtube.com/shorts/NAPi2gWNZP4
Monday, 19 October 2026
Black Hat Europe 2026 | Shaping the Future of AI and Cybersecurity
Join the global cybersecurity community at Black Hat Europe 2026 in London. Explore world-class Trainings, groundbreaking Briefings, hands-on demonstrations, and unparalleled networking opportunities.
Black Hat Europe 2026 brings together security professionals, researchers, engineers, practitioners, and business leaders from around the world for four days of learning, collaboration, and innovation.
This is where the cybersecurity community comes together to exchange knowledge, explore emerging threats, and discover the latest security research.
✔ Hands-on cybersecurity Trainings
✔ Expert-led Briefings and original research
✔ Arsenal demonstrations and cutting-edge tools
✔ Networking opportunities with the global security community
✔ Business Hall showcasing leading security solutions
📍 London, United Kingdom
📅 7-10 December 2026
Register today: https://ift.tt/L5nzW6K
#BHEU #Cybersecurity #InfoSec #SecurityResearch #CyberSecurityTraining #BlackHat
source https://www.youtube.com/watch?v=1PZ8fu1v14Q
source https://www.youtube.com/watch?v=1PZ8fu1v14Q
Subscribe to:
Posts (Atom)
-
WeChat, with over 1.2 billion monthly active users, stands as the most popular messaging and social media platform in China and third global...
-
It's coming, and you aren't ready—your first generative AI chatbot incident. GenAI chatbots, leveraging LLMs, are revolutionizing cu...