Dissecting the Black Box: Circuit-Level Analysis of LLM Vulnerability Detection

Large language models LLMs can detect software vulnerabilities, but how do they actually identify vulnerable code? We address this question using mechanistic interpretability; analyzing the internal computations of a neural network to understand its reasoning process.Using Circuit Tracer on...

Babel: Jailbreaking Safety Attention Via Obfuscation Distribution Optimized Sampling

Despite rigorous safety alignment, Large Language Models LLMs remain vulnerable to jailbreak attacks. Existing black-box methods often rely on heuristic templates or exhaustive trials, lacking mechanistic interpretability and query efficiency. In this study, we investigate an intrinsic...

Astra Linux - уязвимость в linux-5.10, linux-6.1, linux, linux-5.15

In the Linux kernel, the following vulnerability has been resolved: USB: usb-storage: A division-by-zero error has been prevented in the isd200atacommand function. The isd200 sub-driver in usb-storage uses the HEADS and SECTORS values in the ATA ID information to calculate the cylinder and head...

Astra Linux - уязвимость в linux-5.10

In the Linux kernel, the following vulnerability has been resolved: net: skb: Fixed the cross-cache free of KFENCE-alocated skb heads. The value of SKBSMALLHEADCACHESIZE is intentionally set to a non-power-of-2 value e.g., 704 on x8664 to avoid collisions with generic kmalloc bucket sizes. This...

Attention Is Where You Attack

Safety-aligned large language models rely on RLHF and instruction tuning to refuse harmful requests, yet the internal mechanisms implementing safety behavior remain poorly understood. We introduce the Attention Redistribution Attack ARA, a white-box adversarial attack that identifies...

Jailbreaking the Matrix: Nullspace Steering for Controlled Model Subversion

Large language models remain vulnerable to jailbreak attacks -- inputs designed to bypass safety mechanisms and elicit harmful responses -- despite advances in alignment and instruction tuning. We propose Head-Masked Nullspace Steering HMNS, a circuit-level intervention that i identifies attentio...

The Trigger in the Haystack: Extracting and Reconstructing LLM Backdoor Triggers

Detecting whether a model has been poisoned is a longstanding problem in AI security. In this work, we present a practical scanner for identifying sleeper agent-style backdoors in causal language models. Our approach relies on two key findings: first, sleeper agents tend to memorize poisoning dat...

Cyber Threat Detection Enabled by Quantum Computing

Threat detection models in cybersecurity must keep up with shifting traffic, strict feature budgets, and noisy hardware, yet even strong classical systems still miss rare or borderline attacks when the data distribution drifts. Small, near-term quantum processors are now available, but existing...

CVE-2025-20388

In Splunk Enterprise versions below 10.0.1, 9.4.6, 9.3.8, and 9.2.10, and Splunk Cloud Platform versions below 10.1.2507.4, 10.0.2503.7, and 9.3.2411.116, a user who holds a role that contains the high privilege capability changeauthentication could enumerate internal IP addresses and network por...

Siemens SIMATIC Devices Divide By Zero (CVE-2024-27059)

In the Linux kernel, the following vulnerability has been resolved: USB: usb-storage: Prevent divide-by-0 error in isd200atacommand The isd200 sub-driver in usb-storage uses the HEADS and SECTORS values in the ATA ID information to calculate cylinder and head values when creating a CDB for READ o...

EUVD-2013-2240

Malware in sbrugna...

Unity Linux 20.1050a / 20.1060a / 20.1070a Security Update: kernel (UTSA-2025-404875)

The Unity Linux 20 host has a package installed that is affected by a vulnerability as referenced in the UTSA-2025-404875 advisory. In the Linux kernel, the following vulnerability has been resolved: USB: usb-storage: Prevent divide-by-0 error in isd200atacommand The isd200 sub-driver in...

CVE-2023-53337 nilfs2: do not write dirty data after degenerating to read-only

In the Linux kernel, the following vulnerability has been resolved: nilfs2: do not write dirty data after degenerating to read-only According to syzbot's report, markbufferdirty called from nilfssegctordoconstruct outputs a warning with some patterns after nilfs2 detects metadata corruption and...

PT-2025-33682 · Undefined · Undefined

@CISACyber Good catch, @AdamRussellW3! CVE-2025-5494854494948 looks serious—thanks for the heads-up. Patching this one ASAP is a no-brainer. Cybersecurity...

Linux Distros Unpatched Vulnerability : CVE-2024-27059

The Linux/Unix host has one or more packages installed that are impacted by a vulnerability without a vendor supplied patch available. - USB: usb-storage: Prevent divide-by-0 error in isd200atacommand The isd200 sub-driver in usb-storage uses the HEADS and SECTORS values in the ATA ID information...

Mechanistic Interpretability in the Presence of Architectural Obfuscation

Architectural obfuscation - e.g., permuting hidden-state tensors, linearly transforming embedding tables, or remapping tokens - has recently gained traction as a lightweight substitute for heavyweight cryptography in privacy-preserving large-language-model LLM inference. While recent work has sho...

SUSE CVE-2025-21722

In the Linux kernel, the following vulnerability has been resolved: nilfs2: do not force clear folio if buffer is referenced Patch series "nilfs2: protect busy buffer heads from being force-cleared". This series fixes the buffer head state inconsistency issues reported by syzbot that occurs when...

UBUNTU-CVE-2025-21722

In the Linux kernel, the following vulnerability has been resolved: nilfs2: do not force clear folio if buffer is referenced Patch series "nilfs2: protect busy buffer heads from being force-cleared". This series fixes the buffer head state inconsistency issues reported by syzbot that occurs when...

DEBIAN-CVE-2025-21699

In the Linux kernel, the following vulnerability has been resolved: gfs2: Truncate address space when flipping GFS2DIFJDATA flag Truncate an inode's address space when flipping the GFS2DIFJDATA flag: depending on that flag, the pages in the address space will either use buffer heads or...

UBUNTU-CVE-2025-21699

In the Linux kernel, the following vulnerability has been resolved: gfs2: Truncate address space when flipping GFS2DIFJDATA flag Truncate an inode's address space when flipping the GFS2DIFJDATA flag: depending on that flag, the pages in the address space will either use buffer heads or...