Tech-ASan: Two-Stage Check for Address Sanitizer

Address Sanitizer ASan is a sharp weapon for detecting memory safety violations, including temporal and spatial errors hidden in C/C++ programs during execution. However, ASan incurs significant runtime overhead, which limits its efficiency in testing large software. The overhead mainly comes fro...

$PD^3F$: a Pluggable and Dynamic DoS-Defense Framework against Resource Consumption Attacks Targeting Large Language Models

Large Language Models LLMs, due to substantial computational requirements, are vulnerable to resource consumption attacks, which can severely degrade server performance or even cause crashes, as demonstrated by denial-of-service DoS attacks designed for LLMs. However, existing works lack mitigati...

Cryptanalysis of a Lattice-Based PIR Scheme for Arbitrary Database Sizes

Private Information Retrieval PIR schemes enable users to securely retrieve files from a server without disclosing the content of their queries, thereby preserving their privacy. In 2008, Melchor and Gaborit proposed a PIR scheme that achieves a balance between communication overhead and...

Privacy-Preserving Transformers: SwiftKey'S Differential Privacy Implementation

In this paper we train a transformer using differential privacy DP for language modeling in SwiftKey. We run multiple experiments to balance the trade-off between the model size, run-time speed and accuracy. We show that we get small and consistent gains in the next-word-prediction and accuracy...

VIDSTAMP: a Temporally-Aware Watermark for Ownership and Integrity in Video Diffusion Models

The rapid rise of video diffusion models has enabled the generation of highly realistic and temporally coherent videos, raising critical concerns about content authenticity, provenance, and misuse. Existing watermarking approaches, whether passive, post-hoc, or adapted from image-based techniques...

TSCL:Multi-Party Loss Balancing Scheme for Deep Learning Image Steganography Based on Curriculum Learning

For deep learning-based image steganography frameworks, in order to ensure the invisibility and recoverability of the information embedding, the loss function usually contains several losses such as embedding loss, recovery loss and steganalysis loss. In previous research works, fixed loss weight...

MULTI-LF: a Unified Continuous Learning Framework for Real-Time DDoS Detection in Multi-Environment Networks

Detecting Distributed Denial of Service DDoS attacks in Multi-Environment M-En networks presents significant challenges due to diverse malicious traffic patterns and the evolving nature of cyber threats. Existing AI-based detection systems struggle to adapt to new attack strategies and lack...

SUSE CVE-2022-48744

In the Linux kernel, the following vulnerability has been resolved: net/mlx5e: Avoid field-overflowing memcpy In preparation for FORTIFYSOURCE performing compile-time and run-time field bounds checking for memcpy, memmove, and memset, avoid intentionally writing across neighboring fields. Use...

China-Linked Hackers Adopt Two-Stage Infection Tactic to Deploy Deuterbear RAT

Cybersecurity researchers have shed more light on a remote access trojan RAT known as Deuterbear used by the China-linked BlackTech hacking group as part of a cyber espionage campaign targeting the Asia-Pacific region this year. "Deuterbear, while similar to Waterbear in many ways, shows...

New macOS malware RustBucket attributed to North Korean group BlueNoroff

Threat Level Attack Report For a detailed threat advisory, download the pdf file here Summary RustBucket, attributed to BlueNoroff, is split into two stages, with the second stage application appearing as a legitimate PDF viewer but becoming malicious when a specific PDF is loaded. To receive...

Brazilian Banking Trojan Communicates Via Microsoft SQL Server

Researchers have discovered a banking trojan making waves in Brazil with an array of tricks up its sleeve, including using an unusual command and control C&C server and a full-screen social-engineering overlay form. Researchers at IBM X-Force research on Tuesday revealed that attackers are using...

Wannacry depth of analysis: the first stage tasksche-vulnerability warning-the black bar safety net

WannaCry ransomware is a 2017 of the most popular ransomware, which uses a Microsoft vulnerability in the global range attacks make the world more than 100 countries, hundreds of thousands of users by the impact. Has a global range of network security education for all. As a security industry...

Subversion 0.3.7/1.0.0 - Remote Buffer Overflow Exploit

No description provided by source. / hoagiesubversion.c Remote exploit against Subversion-Servers. Author: greuff [email protected] Tested on Subversion 1.0.0 and 0.37 Algorithm: This is a two-stage exploit. The first stage overflows a buffer on the stack and leaves us 60 bytes of machine code to be...

Mercur Messaging 2005 (Windows 2000 SP4) - IMAP 'Subscribe' Remote Overflow

!/usr/bin/python Remote exploit for the stack overflow vulnerability in Mercur Messaging 2005 SP3 IMAP service. The exploit was tested on windows 2000 server SP4 in a Vmware environment. At the time of overflow EBX points to our shellcode. However this buffer into which EBX points will give a...

Microsoft Word Document Code Execution Proof of Concept

No description provided by source. ===== The file I have attached is a very basic two stage bug. stage 1 the first mod forces the code down a wrong path. the second mod by itsself is harmless, however when used with the first it will be the first and part of the second overwrite. I have use...

Microsoft Word Document - Malformed Pointer (PoC)

===== The file I have attached is a very basic two stage bug. stage 1 the first mod forces the code down a wrong path. the second mod by itsself is harmless, however when used with the first it will be the first and part of the second overwrite. I have use 41414141 as a marker to make it easier f...

Subversion 0.3.7/1.0.0 - Remote Buffer Overflow

/ hoagiesubversion.c Remote exploit against Subversion-Servers. Author: greuff Tested on Subversion 1.0.0 and 0.37 Algorithm: This is a two-stage exploit. The first stage overflows a buffer on the stack and leaves us 60 bytes of machine code to be executed. We try to find the socket-fd there and...