Activation Surgery: Jailbreaking White-Box LLMs without Touching the Prompt

Most jailbreak techniques for Large Language Models LLMs primarily rely on prompt modifications, including paraphrasing, obfuscation, or conversational strategies. Meanwhile, abliteration techniques also known as targeted ablations of internal components have been used to study and explain LLM...

Two Frames Matter: A Temporal Attack for Text-To-Video Model Jailbreaking

Recent text-to-video T2V models can synthesize complex videos from lightweight natural language prompts, raising urgent concerns about safety alignment in the event of misuse in the real world. Prior jailbreak attacks typically rewrite unsafe prompts into paraphrases that evade content filters...

LLM Causality Analysis Framework

A comprehensive framework for multi-level causality analysis in Large Language Models LLMs, enabling systematic investigation of safety mechanisms and misbehavior detection across token, neuron, layer, and representation levels. Includes the whitepaper 2512.04841.pdf titled SoK: A Comprehensive...

NeuroBreak: Unveil Internal Jailbreak Mechanisms in Large Language Models

In deployment and application, large language models LLMs typically undergo safety alignment to prevent illegal and unethical outputs. However, the continuous advancement of jailbreak attack techniques, designed to bypass safety mechanisms with adversarial prompts, has placed increasing pressure ...

CVE-2025-7768

Tigo Energy's Cloud Connect Advanced CCA device contains hard-coded credentials that allow unauthorized users to gain administrative access. This vulnerability enables attackers to escalate privileges and take full control of the device, potentially modifying system settings, disrupting solar...

CVE-2025-7768

CVE-2025-7768 affects Tigo Energy Cloud Connect Advanced (CCA). The root cause is hard-coded credentials that grant unauthorized administrative access, enabling privilege escalation and full device control, with potential to modify settings, disrupt solar energy production, and interfere with saf...

Safety Interventions against Adversarial Patches in an Open-Source Driver Assistance System

Drivers are becoming increasingly reliant on advanced driver assistance systems ADAS as autonomous driving technology becomes more popular and developed with advanced safety features to enhance road safety. However, the increasing complexity of the ADAS makes autonomous vehicles AVs more exposed ...

Red-Teaming Text-To-Image Systems by Rule-Based Preference Modeling

Text-to-image T2I models raise ethical and safety concerns due to their potential to generate inappropriate or harmful images. Evaluating these models' security through red-teaming is vital, yet white-box approaches are limited by their need for internal access, complicating their use with...

CVE-2020-18169

A vulnerability in the Windows installer XML WiX toolset of TechSmith Snagit 19.1.1.2860 allows attackers to escalate privileges. NOTE: Exploit of the Snagit installer would require the end user to ignore other safety mechanisms provided by the Host OS. See reference document for more details...

When Safety Detectors Aren'T Enough: a Stealthy and Effective Jailbreak Attack on LLMs Via Steganographic Techniques

Jailbreak attacks pose a serious threat to large language models LLMs by bypassing built-in safety mechanisms and leading to harmful outputs. Studying these attacks is crucial for identifying vulnerabilities and improving model security. This paper presents a systematic survey of jailbreak method...

CVE-2020-18169

A vulnerability in the Windows installer XML WiX toolset of TechSmith Snagit 19.1.1.2860 allows attackers to escalate privileges. NOTE: Exploit of the Snagit installer would require the end user to ignore other safety mechanisms provided by the Host OS. See reference document for more details...

Design/Logic Flaw

A vulnerability in the Windows installer XML WiX toolset of TechSmith Snagit 19.1.1.2860 allows attackers to escalate privileges. NOTE: Exploit of the Snagit installer would require the end user to ignore other safety mechanisms provided by the Host OS. See reference document for more details...