BadReward: Clean-Label Poisoning of Reward Models in Text-To-Image RLHF

Reinforcement Learning from Human Feedback RLHF is crucial for aligning text-to-image T2I models with human preferences. However, RLHF's feedback mechanism also opens new pathways for adversaries. This paper demonstrates the feasibility of hijacking T2I models by poisoning a small fraction of...

Hijacking Large Language Models Via Adversarial In-Context Learning

In-context learning ICL has emerged as a powerful paradigm leveraging LLMs for specific downstream tasks by utilizing labeled examples as demonstrations demos in the preconditioned prompts. Despite its promising performance, crafted adversarial attacks pose a notable threat to the robustness of...

CVE-2022-1839

A vulnerability classified as critical was found in Home Clean Services Management System 1.0. This vulnerability affects the file login.php. The manipulation of the argument email with the input admin%'//AND//SELECT//5383//FROM//SELECTSLEEP2JPeh//AND//'frfq%'='frfq leads to sql injection. The...

CVE-2022-4838

The Clean Login WordPress plugin before 1.13.7 does not validate and escape some of its shortcode attributes before outputting them back in the page, which could allow users with a role as low as contributor to perform Stored Cross-Site Scripting attacks which could be used against high privilege...

CVE-2022-1840

A vulnerability, which was classified as problematic, has been found in Home Clean Services Management System 1.0. This issue affects register.php?link=registerand. The manipulation with the input leads to cross site scripting. The attack may be initiated remotely but demands authentication...

CVE-2022-30052

In Home Clean Service System 1.0, the password parameter is vulnerable to SQL injection attacks...

CVE-2022-1837

A vulnerability was found in Home Clean Services Management System 1.0. It has been rated as critical. Affected by this issue is register.php?link=registerand. The manipulation with the input leads to code execution. The attack may be launched remotely but demands an authentication. Exploit detai...

CVE-2022-1838

A vulnerability classified as critical has been found in Home Clean Services Management System 1.0. This affects an unknown part of admin/login.php. The manipulation of the argument username with the input admin%'//AND//SELECT//5383//FROM//SELECTSLEEP5JPeh//AND//'frfq%'='frfq leads to sql...

CVE-2022-1914

The Clean-Contact WordPress plugin through 1.6 does not have CSRF check in place when updating its settings, which could allow attackers to make a logged in admin change them via a CSRF attack and lead to Stored XSS due to the lack of sanitisation and escaping as well...

CVE-2020-5252

The command-line "safety" package for Python has a potential security issue. There are two Python characteristics that allow malicious code to “poison-pill” command-line Safety package detection routines by disguising, or obfuscating, other malicious or non-secure packages. This vulnerability is...

CVE-2020-6015

Check Point Endpoint Security for Windows before E84.10 can reach denial of service during clean install of the client which will prevent the storage of service log files in non-standard locations...

CVE-2020-14990

IOBit Advanced SystemCare Free 13.5.0.263 allows local users to gain privileges for file deletion by manipulating the Clean & Optimize feature with an NTFS junction and an Object Manager symbolic link...

CVE-2015-9336

The clean-login plugin before 1.5.1 for WordPress has reflected XSS...

CVE-2013-1784

Cross-site scripting XSS vulnerability in the 3 slide gallery in the Clean Theme before 7.x-1.3 for Drupal allows remote authenticated users with the administer themes permission to inject arbitrary web script or HTML via unspecified vectors...

CVE-2005-4825

Cisco Clean Access 3.5.5 and earlier on the Secure Smart Manager allows remote attackers to bypass authentication and cause a denial of service disk consumption, or make unauthorized files accessible, by uploading files through requests to certain JSP scripts, a related issue to CVE-2005-4332...

Reliable Disentanglement Multi-View Learning against View Adversarial Attacks

Trustworthy multi-view learning has attracted extensive attention because evidence learning can provide reliable uncertainty estimation to enhance the credibility of multi-view predictions. Existing trusted multi-view learning methods implicitly assume that multi-view data is secure. However, in...

CVE-2025-37986 usb: typec: class: Invalidate USB device pointers on partner unregistration

In the Linux kernel, the following vulnerability has been resolved: usb: typec: class: Invalidate USB device pointers on partner unregistration To avoid using invalid USB device pointers after a Type-C partner disconnects, this patch clears the pointers upon partner unregistration. This ensures a...

CVE-2025-37986

CVE-2025-37986 — Linux kernel USB Type-C issue : A patch in the kernel fixes a vulnerability in the usb: typec class by invalidating/clearing USB device pointers when a Type-C partner unregisters. The change prevents use of stale pointers after a partner disconnect, ensuring a clean state for fut...

kernel: smb: client: Fix use-after-free of network namespace.

A flaw was found in the Linux kernel's Server Message Block SMB client. Improper reference counting allows kernel sockets to persist after their associated network namespace has been released. A local attacker with permission to create and destroy network namespaces—such as in containerized...

Where the Devil Hides: Deepfake Detectors Can No Longer Be Trusted

With the advancement of AI generative techniques, Deepfake faces have become incredibly realistic and nearly indistinguishable to the human eye. To counter this, Deepfake detectors have been developed as reliable tools for assessing face authenticity. These detectors are typically developed on De...