Abstract:Recent advancements in AI safety have led to increased efforts in training and red-teaming large language models (LLMs) to mitigate unsafe content generation. However, these safety mechanisms may not be comprehensive, leaving potential vulnerabilities unexplored. This paper introduces MathPrompt, a novel jailbreaking technique that exploits LLMs' advanced capabilities in symbolic mathematics to bypass their safety mechanisms. By encoding harmful natural language prompts into mathematical problems, we demonstrate a critical vulnerability in current AI safety measures. Our experiments across 13 state-of-the-art LLMs reveal an average attack success rate of 73.6\%, highlighting the inability of existing safety training mechanisms to generalize to mathematically encoded inputs. Analysis of embedding vectors shows a substantial semantic shift between original and encoded prompts, helping explain the attack's success. This work emphasizes the importance of a holistic approach to AI safety, calling for expanded red-teaming efforts to develop robust safeguards across all potential input types and their associated risks.
Abstract:Event detection and text reasoning have become critical applications across various domains. While LLMs have recently demonstrated impressive progress in reasoning abilities, they often struggle with event detection, particularly due to the absence of training methods that consider causal relationships between event triggers and types. To address this challenge, we propose a novel approach for instruction fine-tuning LLMs for event detection. Our method introduces Semantic Causal Graphs (SCGs) to capture both causal relationships and contextual information within text. Building off of SCGs, we propose SCG Instructions for fine-tuning LLMs by focusing on event triggers and their relationships to event types, and employ Low-Rank Adaptation (LoRA) to help preserve the general reasoning abilities of LLMs. Our evaluations demonstrate that training LLMs with SCG Instructions outperforms standard instruction fine-tuning by an average of 35.69\% on Event Trigger Classification. Notably, our fine-tuned Mistral 7B model also outperforms GPT-4 on key event detection metrics by an average of 31.01\% on Event Trigger Identification, 37.40\% on Event Trigger Classification, and 16.43\% on Event Classification. We analyze the retention of general capabilities, observing only a minimal average drop of 2.03 points across six benchmarks. This comprehensive study investigates multiple LLMs for the event detection task across various datasets, prompting strategies, and training approaches.
Abstract:With the recent proliferation of Large Language Models (LLMs), there has been an increasing demand for tools to detect machine-generated text. The effective detection of machine-generated text face two pertinent problems: First, they are severely limited in generalizing against real-world scenarios, where machine-generated text is produced by a variety of generators, including but not limited to GPT-4 and Dolly, and spans diverse domains, ranging from academic manuscripts to social media posts. Second, existing detection methodologies treat texts produced by LLMs through a restrictive binary classification lens, neglecting the nuanced diversity of artifacts generated by different LLMs. In this work, we undertake a systematic study on the detection of machine-generated text in real-world scenarios. We first study the effectiveness of state-of-the-art approaches and find that they are severely limited against text produced by diverse generators and domains in the real world. Furthermore, t-SNE visualizations of the embeddings from a pretrained LLM's encoder show that they cannot reliably distinguish between human and machine-generated text. Based on our findings, we introduce a novel system, T5LLMCipher, for detecting machine-generated text using a pretrained T5 encoder combined with LLM embedding sub-clustering to address the text produced by diverse generators and domains in the real world. We evaluate our approach across 9 machine-generated text systems and 9 domains and find that our approach provides state-of-the-art generalization ability, with an average increase in F1 score on machine-generated text of 19.6\% on unseen generators and domains compared to the top performing existing approaches and correctly attributes the generator of text with an accuracy of 93.6\%.
Abstract:The critical threat of phishing emails has been further exacerbated by the potential of LLMs to generate highly targeted, personalized, and automated spear phishing attacks. Two critical problems concerning LLM-facilitated phishing require further investigation: 1) Existing studies on lateral phishing lack specific examination of LLM integration for large-scale attacks targeting the entire organization, and 2) Current anti-phishing infrastructure, despite its extensive development, lacks the capability to prevent LLM-generated attacks, potentially impacting both employees and IT security incident management. However, the execution of such investigative studies necessitates a real-world environment, one that functions during regular business operations and mirrors the complexity of a large organizational infrastructure. This setting must also offer the flexibility required to facilitate a diverse array of experimental conditions, particularly the incorporation of phishing emails crafted by LLMs. This study is a pioneering exploration into the use of Large Language Models (LLMs) for the creation of targeted lateral phishing emails, targeting a large tier 1 university's operation and workforce of approximately 9,000 individuals over an 11-month period. It also evaluates the capability of email filtering infrastructure to detect such LLM-generated phishing attempts, providing insights into their effectiveness and identifying potential areas for improvement. Based on our findings, we propose machine learning-based detection techniques for such emails to detect LLM-generated phishing emails that were missed by the existing infrastructure, with an F1-score of 98.96.