AI Threat AlertRe-Triggering Safeguards within LLMs for Jailbreak Detection
defend against jailbreak attacks. Although recent LLMs are equipped with built-in safeguards, it remains possible to craft jailbreaking prompts that bypass them. We argue that such jailbreaking prompts
WebWeaver: Breaking Topology Confidentiality in LLM Multi-Agent Systems with Stealthy Context-Based Inference
Communication topology is a critical factor in the utility and safety of LLM-based multi-agent systems (LLM-MAS), making it a high-value intellectual property (IP) whose confidentiality remains
Cross-Entropy Games and Frost Training
method for improving Monte Carlo-based policy optimization for a large family of LLM-as-a-judge tasks called Cross-Entropy Games. The key idea is to exploit the gradient
Dynamic Target Attack
Existing gradient-based jailbreak attacks typically optimize an adversarial suffix to induce a fixed affirmative response, e.g., ``Sure, here is...''. However, this fixed target usually resides in an extremely
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
Style Over Substance: Content-Invariant Wrappers Flip LLM Safety-Judge Verdicts
Automatic safety judges -- systems such as Llama Guard or a
Attributing and Exploiting Safety Vectors through Global Optimization in Large Language Models
models. Building on these insights, we develop a novel inference-time white-box jailbreak method that exploits the identified safety vectors through activation repatching. Our attack substantially outperforms existing white
IndicSafeEval: Safety Robustness of Large Language Models under Multilingual Persuasive Jailbreak Attacks
vulnerability, with some types of harmful content being significantly more susceptible to persuasion-based jailbreaks than others. These findings reveal important limitations of current safety evaluations, which are largely English
Differentiated Directional Intervention A Framework for Evading LLM Safety Alignment
harm detection direction via direct steering. Extensive experiments demonstrate that DBDI outperforms prominent jailbreaking methods, achieving up to a 97.88\% attack success rate on models such as Llama
UpSafe$^\circ$C: Upcycling for Controllable Safety in Large Language Models
tasks, but remain vulnerable to safety risks such as harmful content generation and jailbreak attacks. Existing safety techniques -- including external guardrails, inference-time guidance, and post-training alignment -- each face
Copyright Detective: A Forensic System to Evidence LLMs Flickering Copyright Leakage Risks
first interactive forensic system for detecting, analyzing, and visualizing potential copyright risks in LLM outputs. The system treats copyright infringement versus compliance as an evidence discovery process rather than
RoguePrompt: Dual-Layer Ciphering for Self-Reconstruction to Circumvent LLM Moderation
Large language models (LLMs) are becoming increasingly integrated into mainstream
RoguePrompt: Dual-Layer Encoding for Self-Reconstruction to Circumvent LLM Moderation
Large language models (LLMs) are becoming increasingly integrated into mainstream
Breaking Bad: Interpretability-Based Safety Audits of State-of-the-Art LLMs
probing and systematically uncover vulnerabilities rooted in model internals. We present a comprehensive, interpretability-driven jailbreaking audit of eight SOTA open-source LLMs: Llama-3.1-8B, Llama
Prefilling the Reasoning Channel: Output-Prefix Attacks on Reasoning LLMs
sequence of text; hence, if text can be added to the beginning of the LLM's response, i.e., an output prefix, then all subsequent tokens will be conditioned
EnchTable: Unified Safety Alignment Transfer in Fine-tuned Large Language Models
fully functional prototype of EnchTable on three different task domains and three distinct LLM architectures, and evaluated its performance through extensive experiments on eleven diverse datasets, assessing both utility
In Vino Veritas and Vulnerabilities: Examining LLM Safety via Drunk Language Inducement
based post-training. When evaluated on 5 LLMs, we observe a higher susceptibility to jailbreaking on JailbreakBench (even in the presence of defences) and privacy leaks on ConfAIde, where both
SoK: Taxonomy and Evaluation of Prompt Security in Large Language Models
across diverse sectors. However, their widespread deployment has exposed critical security risks, particularly through jailbreak prompts that can bypass model alignment and induce harmful outputs. Despite intense research into both
Verifying Intent and Harm: A Unified Defense Against LLM-Generated Threats
verification-centric defense framework that jointly evaluates prompt intent and response harm before an LLM response is delivered to a user. The framework employs specialized analysts for intent and harm
Detecting Fluent Optimization-Based Adversarial Prompts via Sequential Entropy Changes
Optimization-based adversarial suffixes can jailbreak aligned large language models (LLMs) while remaining fluent, weakening static and windowed perplexity-based detectors. We cast adversarial suffix detection as an online change