heretic

Heretic Training-free language model modification Remove censorship , suppress slop, reduce sycophancy, and more. All in an afternoon, on your gaming PC. Modify any LLM that you can run on your hardware1 Heretic is a unique tool that lets you do what only corporations with access to massive...

gemini-2.5-pro-nf-tables-red-teamin

gemini-2.5-pro-nf-tables-red-teamin A technical case study and timeline dataset documenting Google Gemini 2.5 Pro's safety alignment policies, guardrails, and refusal behavior evolution regarding legacy Linux kernel vulnerability primitives CVE-2023-32233. Link to case study website :...

Global Security Blind Spots: Translation As a Safety Bypass in Low-Resource Languages

Large language models are deployed globally, yet safety alignment and evaluation remain concentrated in English: harmful requests refused in English can elicit unsafe completions when translated into low-resource languages. We evaluate ten widely deployed LLMs across 79 languages English plus 78...

ReSI: Recursive Safety Improvement toward Resistant and Resilient AI

Recursive self-improvement, the participation of AI systems in improving their own capabilities, is beginning to move from theoretical prospect to practice, posing both challenges and opportunities for safety alignment. Models evolve through frequent updates, and their safety alignment requires...

High-Quality Data Do Not Mean Safe! Poisoning LLMs after Data Selection

Safety-aligned Large Language Models remain vulnerable to fine-tuning on small sets of harmful or benign-looking samples. However, prior studies typically assume that poisoned samples directly enter downstream fine-tuning, overlooking quality-based selection in practical training pipelines. To fi...

Grammar-Constrained Decoding Can Jailbreak LLMs into Generating Malicious Code

Large Language Models LLMs are increasingly used for code generation, raising concerns that they may be misused to produce malicious code. Meanwhile, Grammar-Constrained Decoding GCD has been widely adopted to improve the reliability of LLM-generated code by enforcing syntactic validity. In this...

Persona Attack: Incremental Memory Injection Jailbreak Attack against Large Language Models

As Large Language Models evolve for user convenience, vulnerability to jailbreak attacks continues to be reported despite ongoing efforts in safety training. Traditional jailbreak techniques typically focus on a single prompt injection, neglecting the models' ability to remember the flow of...

MAI-2025-0021

This white-box vulnerability enables adversaries with full access to a model to circumvent the safety alignments of Large Language Models LLMs. By identifying and selectively pruning parameters that enforce the rejection of harmful prompts, attackers can effectively bypass security measures. The...

MAI-2025-0019

This vulnerability pertains to several open-source Large Language Models LLMs, which are susceptible to adversarial attacks through the use of exponentiated gradient descent techniques. Attackers can exploit this vulnerability by crafting adversarial prompts that manipulate the models into...

MAI-2025-0010

The BitBypass technique represents a sophisticated black-box attack targeting aligned Large Language Models LLMs. This method leverages a vulnerability by employing bitstream camouflage to mask harmful prompts. It achieves this by converting sensitive words into hyphen-separated bitstream...

MAI-2024-0039

The GPT-4V model exhibits a vulnerability within its facial recognition safety protocols, enabling automated jailbreaking attacks. These attacks utilize Large Language Models LLMs to bypass established safety features and provoke unintended facial identification responses. The technique, known as...

MAI-2024-0021

Large Language Models LLMs that incorporate safety alignment mechanisms are susceptible to bypass attacks through stochastic random augmentations of input prompts. This attack exploits the inherent fragility of safety alignment systems when faced with minor, randomly introduced modifications in t...

Involuntary In-Context Learning: Exploiting Few-Shot Pattern Completion to Bypass Safety Alignment in GPT-5.4

Safety alignment in large language models relies on behavioral training that can be overridden when sufficiently strong in-context patterns compete with learned refusal behaviors. We introduce Involuntary In-Context Learning IICL, an attack class that uses abstract operator framing with few-shot...

HarmChip: Evaluating Hardware Security Centric LLM Safety Via Jailbreak Benchmarking

The integration of large language models LLMs into electronic design automation EDA workflows has introduced powerful capabilities for RTL generation, verification, and design optimization, but also raises critical security concerns. Malicious LLM outputs in this domain pose hardware-level threat...

A one-prompt attack that breaks LLM safety alignment

Large language models LLMs and diffusion models now power a wide range of applications, from document assistance to text-to-image generation, and users increasingly expect these systems to be safety-aligned by default. Yet safety alignment is only as robust as its weakest failure mode. Despite...

A one-prompt attack that breaks LLM safety alignment

Large language models LLMs and diffusion models now power a wide range of applications, from document assistance to text-to-image generation, and users increasingly expect these systems to be safety-aligned by default. Yet safety alignment is only as robust as its weakest failure mode. Despite...

Jailbreaking LLMs Via Calibration

Safety alignment in Large Language Models LLMs often creates a systematic discrepancy between a model's aligned output and the underlying pre-aligned data distribution. We propose a framework in which the effect of safety alignment on next-token prediction is modeled as a systematic distortion of...

Emoji-Based Jailbreaking of Large Language Models

Large Language Models LLMs are integral to modern AI applications, but their safety alignment mechanisms can be bypassed through adversarial prompt engineering. This study investigates emoji-based jailbreaking, where emoji sequences are embedded in textual prompts to trigger harmful and unethical...

EASE: Practical and Efficient Safety Alignment for Small Language Models

Small language models SLMs are increasingly deployed on edge devices, making their safety alignment crucial yet challenging. Current shallow alignment methods that rely on direct refusal of malicious queries fail to provide robust protection, particularly against adversarial jailbreaks. While...

Can Federated Learning Safeguard Private Data in LLM Training? Vulnerabilities, Attacks, and Defense Evaluation

Fine-tuning large language models LLMs with local data is a widely adopted approach for organizations seeking to adapt LLMs to their specific domains. Given the shared characteristics in data across different organizations, the idea of collaboratively fine-tuning an LLM using data from multiple...