Search

Large language models (LLMs) have exhibited remarkable capabilities in natural language generation, but they have also been observed to magnify societal biases, particularly those related to gender. In response to this issue, several benchmarks have been proposed to assess gender bias in LLMs. However, these benchmarks often lack practical flexibility or inadvertently introduce biases. To address these shortcomings, we introduce GenderCARE, a comprehensive framework that encompasses innovative Criteria, bias Assessment, Reduction techniques, and Evaluation metrics for quantifying and mitigating gender bias in LLMs. To begin, we establish pioneering criteria for gender equality benchmarks, spanning dimensions such as inclusivity, diversity, explainability, objectivity, robustness, and realisticity. Guided by these criteria, we construct GenderPair, a novel pair-based benchmark designed to assess gender bias in LLMs comprehensively. Our benchmark provides standardized and realistic evaluations, including previously overlooked gender groups such as transgender and non-binary individuals. Furthermore, we develop effective debiasing techniques that incorporate counterfactual data augmentation and specialized fine-tuning strategies to reduce gender bias in LLMs without compromising their overall performance. Extensive experiments demonstrate a significant reduction in various gender bias benchmarks, with reductions peaking at over 90% and averaging above 35% across 17 different LLMs. Importantly, these reductions come with minimal variability in mainstream language tasks, remaining below 2%. By offering a realistic assessment and tailored reduction of gender biases, we hope that our GenderCARE can represent a significant step towards achieving fairness and equity in LLMs. More details are available at https://github.com/kstanghere/GenderCARE-ccs24.

Large language models (LLMs) like ChatGPT and Gemini have significantly advanced natural language processing, enabling various applications such as chatbots and automated content generation. However, these models can be exploited by malicious individuals who craft toxic prompts to elicit harmful or unethical responses. These individuals often employ jailbreaking techniques to bypass safety mechanisms, highlighting the need for robust toxic prompt detection methods. Existing detection techniques, both blackbox and whitebox, face challenges related to the diversity of toxic prompts, scalability, and computational efficiency. In response, we propose ToxicDetector, a lightweight greybox method designed to efficiently detect toxic prompts in LLMs. ToxicDetector leverages LLMs to create toxic concept prompts, uses embedding vectors to form feature vectors, and employs a Multi-Layer Perceptron (MLP) classifier for prompt classification. Our evaluation on various versions of the LLama models, Gemma-2, and multiple datasets demonstrates that ToxicDetector achieves a high accuracy of 96.39\% and a low false positive rate of 2.00\%, outperforming state-of-the-art methods. Additionally, ToxicDetector's processing time of 0.0780 seconds per prompt makes it highly suitable for real-time applications. ToxicDetector achieves high accuracy, efficiency, and scalability, making it a practical method for toxic prompt detection in LLMs., Comment: Accepted by the 39th IEEE/ACM International Conference on Automated Software Engineering (ASE 2024)

Geolocation is now a vital aspect of modern life, offering numerous benefits but also presenting serious privacy concerns. The advent of large vision-language models (LVLMs) with advanced image-processing capabilities introduces new risks, as these models can inadvertently reveal sensitive geolocation information. This paper presents the first in-depth study analyzing the challenges posed by traditional deep learning and LVLM-based geolocation methods. Our findings reveal that LVLMs can accurately determine geolocations from images, even without explicit geographic training. To address these challenges, we introduce \tool{}, an innovative framework that significantly enhances image-based geolocation accuracy. \tool{} employs a systematic chain-of-thought (CoT) approach, mimicking human geoguessing strategies by carefully analyzing visual and contextual cues such as vehicle types, architectural styles, natural landscapes, and cultural elements. Extensive testing on a dataset of 50,000 ground-truth data points shows that \tool{} outperforms both traditional models and human benchmarks in accuracy. It achieves an impressive average score of 4550.5 in the GeoGuessr game, with an 85.37\% win rate, and delivers highly precise geolocation predictions, with the closest distances as accurate as 0.3 km. Furthermore, our study highlights issues related to dataset integrity, leading to the creation of a more robust dataset and a refined framework that leverages LVLMs' cognitive capabilities to improve geolocation precision. These findings underscore \tool{}'s superior ability to interpret complex visual data, the urgent need to address emerging security vulnerabilities posed by LVLMs, and the importance of responsible AI development to ensure user privacy protection.

Image captioning (IC) systems, such as Microsoft Azure Cognitive Service, translate image content into descriptive language but can generate inaccuracies leading to misinterpretations. Advanced testing techniques like MetaIC and ROME aim to address these issues but face significant challenges. These methods require intensive manual labor for detailed annotations and often produce unrealistic images, either by adding unrelated objects or failing to remove existing ones. Additionally, they generate limited test suites, with MetaIC restricted to inserting specific objects and ROME limited to a narrow range of variations. We introduce SPOLRE, a novel automated tool for semantic-preserving object layout reconstruction in IC system testing. SPOLRE leverages four transformation techniques to modify object layouts without altering the image's semantics. This automated approach eliminates the need for manual annotations and creates realistic, varied test suites. Our tests show that over 75% of survey respondents find SPOLRE-generated images more realistic than those from state-of-the-art methods. SPOLRE excels in identifying caption errors, detecting 31,544 incorrect captions across seven IC systems with an average precision of 91.62%, surpassing other methods which average 85.65% accuracy and identify 17,160 incorrect captions. Notably, SPOLRE identified 6,236 unique issues within Azure, demonstrating its effectiveness against one of the most advanced IC systems.

Timely stress detection is crucial for protecting vulnerable groups from long-term detrimental effects by enabling early intervention. Wearable devices, by collecting real-time physiological signals, offer a solution for accurate stress detection accommodating individual differences. This position paper introduces an adaptive framework for personalized stress detection using PPG and EDA signals. Unlike traditional methods that rely on a generalized model, which may suffer performance drops when applied to new users due to domain shifts, this framework aims to provide each user with a personalized model for higher stress detection accuracy. The framework involves three stages: developing a generalized model offline with an initial dataset, adapting the model to the user's unlabeled data, and fine-tuning it with a small set of labeled data obtained through user interaction. This approach not only offers a foundation for mobile applications that provide personalized stress detection and intervention but also has the potential to address a wider range of mental health issues beyond stress detection using physiological signals.

Security concerns for large language models (LLMs) have recently escalated, focusing on thwarting jailbreaking attempts in discrete prompts. However, the exploration of jailbreak vulnerabilities arising from continuous embeddings has been limited, as prior approaches primarily involved appending discrete or continuous suffixes to inputs. Our study presents a novel channel for conducting direct attacks on LLM inputs, eliminating the need for suffix addition or specific questions provided that the desired output is predefined. We additionally observe that extensive iterations often lead to overfitting, characterized by repetition in the output. To counteract this, we propose a simple yet effective strategy named CLIP. Our experiments show that for an input length of 40 at iteration 1000, applying CLIP improves the ASR from 62% to 83%

To support software developers in understanding and maintaining programs, various automatic (source) code summarization techniques have been proposed to generate a concise natural language summary (i.e., comment) for a given code snippet. Recently, the emergence of large language models (LLMs) has led to a great boost in the performance of code-related tasks. In this paper, we undertake a systematic and comprehensive study on code summarization in the era of LLMs, which covers multiple aspects involved in the workflow of LLM-based code summarization. Specifically, we begin by examining prevalent automated evaluation methods for assessing the quality of summaries generated by LLMs and find that the results of the GPT-4 evaluation method are most closely aligned with human evaluation. Then, we explore the effectiveness of five prompting techniques (zero-shot, few-shot, chain-of-thought, critique, and expert) in adapting LLMs to code summarization tasks. Contrary to expectations, advanced prompting techniques may not outperform simple zero-shot prompting. Next, we investigate the impact of LLMs' model settings (including top\_p and temperature parameters) on the quality of generated summaries. We find the impact of the two parameters on summary quality varies by the base LLM and programming language, but their impacts are similar. Moreover, we canvass LLMs' abilities to summarize code snippets in distinct types of programming languages. The results reveal that LLMs perform suboptimally when summarizing code written in logic programming languages compared to other language types. Finally, we unexpectedly find that CodeLlama-Instruct with 7B parameters can outperform advanced GPT-4 in generating summaries describing code implementation details and asserting code properties. We hope that our findings can provide a comprehensive understanding of code summarization in the era of LLMs., Comment: Just accepted to the 47th International Conference on Software Engineering (ICSE 2025)

Upon deployment to edge devices, it is often desirable for a model to further learn from streaming data to improve accuracy. However, extracting representative features from such data is challenging because it is typically unlabeled, non-independent and identically distributed (non-i.i.d), and is seen only once. To mitigate this issue, a common strategy is to maintain a small data buffer on the edge device to hold the most representative data for further learning. As most data is either never stored or quickly discarded, identifying the most representative data to avoid significant information loss becomes critical. In this paper, we propose an on-device framework that addresses this issue by condensing incoming data into more informative samples. Specifically, to effectively handle unlabeled incoming data, we propose a pseudo-labeling technique designed for unlabeled on-device learning environments. Additionally, we develop a dataset condensation technique that only requires little computation resources. To counteract the effects of noisy labels during the condensation process, we further utilize a contrastive learning objective to improve the purity of class data within the buffer. Our empirical results indicate substantial improvements over existing methods, particularly when buffer capacity is severely restricted. For instance, with a buffer capacity of just one sample per class, our method achieves an accuracy that outperforms the best existing baseline by 58.4% on the CIFAR-10 dataset., Comment: 9 pages, 10 figures

Large language models (LLMs) have transformed the field of natural language processing, but they remain susceptible to jailbreaking attacks that exploit their capabilities to generate unintended and potentially harmful content. Existing token-level jailbreaking techniques, while effective, face scalability and efficiency challenges, especially as models undergo frequent updates and incorporate advanced defensive measures. In this paper, we introduce JailMine, an innovative token-level manipulation approach that addresses these limitations effectively. JailMine employs an automated "mining" process to elicit malicious responses from LLMs by strategically selecting affirmative outputs and iteratively reducing the likelihood of rejection. Through rigorous testing across multiple well-known LLMs and datasets, we demonstrate JailMine's effectiveness and efficiency, achieving a significant average reduction of 86% in time consumed while maintaining high success rates averaging 95%, even in the face of evolving defensive strategies. Our work contributes to the ongoing effort to assess and mitigate the vulnerability of LLMs to jailbreaking attacks, underscoring the importance of continued vigilance and proactive measures to enhance the security and reliability of these powerful language models.

CAPTCHAs have become a ubiquitous tool in safeguarding applications from automated bots. Over time, the arms race between CAPTCHA development and evasion techniques has led to increasingly sophisticated and diverse designs. The latest iteration, reasoning CAPTCHAs, exploits tasks that are intuitively simple for humans but challenging for conventional AI technologies, thereby enhancing security measures. Driven by the evolving AI capabilities, particularly the advancements in Large Language Models (LLMs), we investigate the potential of multimodal LLMs to solve modern reasoning CAPTCHAs. Our empirical analysis reveals that, despite their advanced reasoning capabilities, LLMs struggle to solve these CAPTCHAs effectively. In response, we introduce Oedipus, an innovative end-to-end framework for automated reasoning CAPTCHA solving. Central to this framework is a novel strategy that dissects the complex and human-easy-AI-hard tasks into a sequence of simpler and AI-easy steps. This is achieved through the development of a Domain Specific Language (DSL) for CAPTCHAs that guides LLMs in generating actionable sub-steps for each CAPTCHA challenge. The DSL is customized to ensure that each unit operation is a highly solvable subtask revealed in our previous empirical study. These sub-steps are then tackled sequentially using the Chain-of-Thought (CoT) methodology. Our evaluation shows that Oedipus effectively resolves the studied CAPTCHAs, achieving an average success rate of 63.5\%. Remarkably, it also shows adaptability to the most recent CAPTCHA designs introduced in late 2023, which are not included in our initial study. This prompts a discussion on future strategies for designing reasoning CAPTCHAs that can effectively counter advanced AI solutions.

With the expanding application of Large Language Models (LLMs) in various domains, it becomes imperative to comprehensively investigate their unforeseen behaviors and consequent outcomes. In this study, we introduce and systematically explore the phenomenon of "glitch tokens", which are anomalous tokens produced by established tokenizers and could potentially compromise the models' quality of response. Specifically, we experiment on seven top popular LLMs utilizing three distinct tokenizers and involving a totally of 182,517 tokens. We present categorizations of the identified glitch tokens and symptoms exhibited by LLMs when interacting with glitch tokens. Based on our observation that glitch tokens tend to cluster in the embedding space, we propose GlitchHunter, a novel iterative clustering-based technique, for efficient glitch token detection. The evaluation shows that our approach notably outperforms three baseline methods on eight open-source LLMs. To the best of our knowledge, we present the first comprehensive study on glitch tokens. Our new detection further provides valuable insights into mitigating tokenization-related errors in LLMs.

Augmented generation techniques such as Retrieval-Augmented Generation (RAG) and Cache-Augmented Generation (CAG) have revolutionized the field by enhancing large language model (LLM) outputs with external knowledge and cached information. However, the integration of vector databases, which serve as a backbone for these augmentations, introduces critical challenges, particularly in ensuring accurate vector matching. False vector matching in these databases can significantly compromise the integrity and reliability of LLM outputs, leading to misinformation or erroneous responses. Despite the crucial impact of these issues, there is a notable research gap in methods to effectively detect and address false vector matches in LLM-augmented generation. This paper presents MeTMaP, a metamorphic testing framework developed to identify false vector matching in LLM-augmented generation systems. We derive eight metamorphic relations (MRs) from six NLP datasets, which form our method's core, based on the idea that semantically similar texts should match and dissimilar ones should not. MeTMaP uses these MRs to create sentence triplets for testing, simulating real-world LLM scenarios. Our evaluation of MeTMaP over 203 vector matching configurations, involving 29 embedding models and 7 distance metrics, uncovers significant inaccuracies. The results, showing a maximum accuracy of only 41.51\% on our tests compared to the original datasets, emphasize the widespread issue of false matches in vector matching methods and the critical need for effective detection and mitigation in LLM-augmented applications.

Large Language Models (LLMS) have increasingly become central to generating content with potential societal impacts. Notably, these models have demonstrated capabilities for generating content that could be deemed harmful. To mitigate these risks, researchers have adopted safety training techniques to align model outputs with societal values to curb the generation of malicious content. However, the phenomenon of "jailbreaking", where carefully crafted prompts elicit harmful responses from models, persists as a significant challenge. This research conducts a comprehensive analysis of existing studies on jailbreaking LLMs and their defense techniques. We meticulously investigate nine attack techniques and seven defense techniques applied across three distinct language models: Vicuna, LLama, and GPT-3.5 Turbo. We aim to evaluate the effectiveness of these attack and defense techniques. Our findings reveal that existing white-box attacks underperform compared to universal techniques and that including special tokens in the input significantly affects the likelihood of successful attacks. This research highlights the need to concentrate on the security facets of LLMs. Additionally, we contribute to the field by releasing our datasets and testing framework, aiming to foster further research into LLM security. We believe these contributions will facilitate the exploration of security measures within this domain., Comment: 18 pages, 9 figures, Accepted in ACL 2024

With the prevalence of text-to-image generative models, their safety becomes a critical concern. adversarial testing techniques have been developed to probe whether such models can be prompted to produce Not-Safe-For-Work (NSFW) content. However, existing solutions face several challenges, including low success rate and inefficiency. We introduce Groot, the first automated framework leveraging tree-based semantic transformation for adversarial testing of text-to-image models. Groot employs semantic decomposition and sensitive element drowning strategies in conjunction with LLMs to systematically refine adversarial prompts. Our comprehensive evaluation confirms the efficacy of Groot, which not only exceeds the performance of current state-of-the-art approaches but also achieves a remarkable success rate (93.66%) on leading text-to-image models such as DALL-E 3 and Midjourney.

Large Language Models~(LLMs) have gained immense popularity and are being increasingly applied in various domains. Consequently, ensuring the security of these models is of paramount importance. Jailbreak attacks, which manipulate LLMs to generate malicious content, are recognized as a significant vulnerability. While existing research has predominantly focused on direct jailbreak attacks on LLMs, there has been limited exploration of indirect methods. The integration of various plugins into LLMs, notably Retrieval Augmented Generation~(RAG), which enables LLMs to incorporate external knowledge bases into their response generation such as GPTs, introduces new avenues for indirect jailbreak attacks. To fill this gap, we investigate indirect jailbreak attacks on LLMs, particularly GPTs, introducing a novel attack vector named Retrieval Augmented Generation Poisoning. This method, Pandora, exploits the synergy between LLMs and RAG through prompt manipulation to generate unexpected responses. Pandora uses maliciously crafted content to influence the RAG process, effectively initiating jailbreak attacks. Our preliminary tests show that Pandora successfully conducts jailbreak attacks in four different scenarios, achieving higher success rates than direct attacks, with 64.3\% for GPT-3.5 and 34.8\% for GPT-4., Comment: 6 pages

The advent of MiniApps, operating within larger SuperApps, has revolutionized user experiences by offering a wide range of services without the need for individual app downloads. However, this convenience has raised significant privacy concerns, as these MiniApps often require access to sensitive data, potentially leading to privacy violations. Our research addresses the critical gaps in the analysis of MiniApps' privacy practices, especially focusing on WeChat MiniApps in the Android ecosystem. Despite existing privacy regulations and platform guidelines, there is a lack of effective mechanisms to safeguard user privacy fully. We introduce MiniScope, a novel two-phase hybrid analysis approach, specifically designed for the MiniApp environment. This approach overcomes the limitations of existing static analysis techniques by incorporating dynamic UI exploration for complete code coverage and accurate privacy practice identification. Our methodology includes modeling UI transition states, resolving cross-package callback control flows, and automated iterative UI exploration. This allows for a comprehensive understanding of MiniApps' privacy practices, addressing the unique challenges of sub-package loading and event-driven callbacks. Our empirical evaluation of over 120K MiniApps using MiniScope demonstrates its effectiveness in identifying privacy inconsistencies. The results reveal significant issues, with 5.7% of MiniApps over-collecting private data and 33.4% overclaiming data collection. These findings emphasize the urgent need for more precise privacy monitoring systems and highlight the responsibility of SuperApp operators to enforce stricter privacy measures.

Pre-training, which utilizes extensive and varied datasets, is a critical factor in the success of Large Language Models (LLMs) across numerous applications. However, the detailed makeup of these datasets is often not disclosed, leading to concerns about data security and potential misuse. This is particularly relevant when copyrighted material, still under legal protection, is used inappropriately, either intentionally or unintentionally, infringing on the rights of the authors. In this paper, we introduce a detailed framework designed to detect and assess the presence of content from potentially copyrighted books within the training datasets of LLMs. This framework also provides a confidence estimation for the likelihood of each content sample's inclusion. To validate our approach, we conduct a series of simulated experiments, the results of which affirm the framework's effectiveness in identifying and addressing instances of content misuse in LLM training processes. Furthermore, we investigate the presence of recognizable quotes from famous literary works within these datasets. The outcomes of our study have significant implications for ensuring the ethical use of copyrighted materials in the development of LLMs, highlighting the need for more transparent and responsible data management practices in this field.

The popularity of automatic speech recognition (ASR) systems nowadays leads to an increasing need for improving their accessibility. Handling stuttering speech is an important feature for accessible ASR systems. To improve the accessibility of ASR systems for stutterers, we need to expose and analyze the failures of ASR systems on stuttering speech. The speech datasets recorded from stutterers are not diverse enough to expose most of the failures. Furthermore, these datasets lack ground truth information about the non-stuttered text, rendering them unsuitable as comprehensive test suites. Therefore, a methodology for generating stuttering speech as test inputs to test and analyze the performance of ASR systems is needed. However, generating valid test inputs in this scenario is challenging. The reason is that although the generated test inputs should mimic how stutterers speak, they should also be diverse enough to trigger more failures. To address the challenge, we propose ASTER, a technique for automatically testing the accessibility of ASR systems. ASTER can generate valid test cases by injecting five different types of stuttering. The generated test cases can both simulate realistic stuttering speech and expose failures in ASR systems. Moreover, ASTER can further enhance the quality of the test cases with a multi-objective optimization-based seed updating algorithm. We implemented ASTER as a framework and evaluated it on four open-source ASR models and three commercial ASR systems. We conduct a comprehensive evaluation of ASTER and find that it significantly increases the word error rate, match error rate, and word information loss in the evaluated ASR systems. Additionally, our user study demonstrates that the generated stuttering audio is indistinguishable from real-world stuttering audio clips.

Penetration testing, a crucial industrial practice for ensuring system security, has traditionally resisted automation due to the extensive expertise required by human professionals. Large Language Models (LLMs) have shown significant advancements in various domains, and their emergent abilities suggest their potential to revolutionize industries. In this research, we evaluate the performance of LLMs on real-world penetration testing tasks using a robust benchmark created from test machines with platforms. Our findings reveal that while LLMs demonstrate proficiency in specific sub-tasks within the penetration testing process, such as using testing tools, interpreting outputs, and proposing subsequent actions, they also encounter difficulties maintaining an integrated understanding of the overall testing scenario. In response to these insights, we introduce PentestGPT, an LLM-empowered automatic penetration testing tool that leverages the abundant domain knowledge inherent in LLMs. PentestGPT is meticulously designed with three self-interacting modules, each addressing individual sub-tasks of penetration testing, to mitigate the challenges related to context loss. Our evaluation shows that PentestGPT not only outperforms LLMs with a task-completion increase of 228.6\% compared to the \gptthree model among the benchmark targets but also proves effective in tackling real-world penetration testing challenges. Having been open-sourced on GitHub, PentestGPT has garnered over 4,700 stars and fostered active community engagement, attesting to its value and impact in both the academic and industrial spheres.

This paper addresses the prevalent lack of tools to facilitate and empower Game Theory and Artificial Intelligence (AI) research in cybersecurity. The primary contribution is the introduction of ExploitFlow (EF), an AI and Game Theory-driven modular library designed for cyber security exploitation. EF aims to automate attacks, combining exploits from various sources, and capturing system states post-action to reason about them and understand potential attack trees. The motivation behind EF is to bolster Game Theory and AI research in cybersecurity, with robotics as the initial focus. Results indicate that EF is effective for exploring machine learning in robot cybersecurity. An artificial agent powered by EF, using Reinforcement Learning, outperformed both brute-force and human expert approaches, laying the path for using ExploitFlow for further research. Nonetheless, we identified several limitations in EF-driven agents, including a propensity to overfit, the scarcity and production cost of datasets for generalization, and challenges in interpreting networking states across varied security settings. To leverage the strengths of ExploitFlow while addressing identified shortcomings, we present Malism, our vision for a comprehensive automated penetration testing framework with ExploitFlow at its core.

Large Language Models (LLMs) have revolutionized Artificial Intelligence (AI) services due to their exceptional proficiency in understanding and generating human-like text. LLM chatbots, in particular, have seen widespread adoption, transforming human-machine interactions. However, these LLM chatbots are susceptible to "jailbreak" attacks, where malicious users manipulate prompts to elicit inappropriate or sensitive responses, contravening service policies. Despite existing attempts to mitigate such threats, our research reveals a substantial gap in our understanding of these vulnerabilities, largely due to the undisclosed defensive measures implemented by LLM service providers. In this paper, we present Jailbreaker, a comprehensive framework that offers an in-depth understanding of jailbreak attacks and countermeasures. Our work makes a dual contribution. First, we propose an innovative methodology inspired by time-based SQL injection techniques to reverse-engineer the defensive strategies of prominent LLM chatbots, such as ChatGPT, Bard, and Bing Chat. This time-sensitive approach uncovers intricate details about these services' defenses, facilitating a proof-of-concept attack that successfully bypasses their mechanisms. Second, we introduce an automatic generation method for jailbreak prompts. Leveraging a fine-tuned LLM, we validate the potential of automated jailbreak generation across various commercial LLM chatbots. Our method achieves a promising average success rate of 21.58%, significantly outperforming the effectiveness of existing techniques. We have responsibly disclosed our findings to the concerned service providers, underscoring the urgent need for more robust defenses. Jailbreaker thus marks a significant step towards understanding and mitigating jailbreak threats in the realm of LLM chatbots.

Large Language Models (LLMs), renowned for their superior proficiency in language comprehension and generation, stimulate a vibrant ecosystem of applications around them. However, their extensive assimilation into various services introduces significant security risks. This study deconstructs the complexities and implications of prompt injection attacks on actual LLM-integrated applications. Initially, we conduct an exploratory analysis on ten commercial applications, highlighting the constraints of current attack strategies in practice. Prompted by these limitations, we subsequently formulate HouYi, a novel black-box prompt injection attack technique, which draws inspiration from traditional web injection attacks. HouYi is compartmentalized into three crucial elements: a seamlessly-incorporated pre-constructed prompt, an injection prompt inducing context partition, and a malicious payload designed to fulfill the attack objectives. Leveraging HouYi, we unveil previously unknown and severe attack outcomes, such as unrestricted arbitrary LLM usage and uncomplicated application prompt theft. We deploy HouYi on 36 actual LLM-integrated applications and discern 31 applications susceptible to prompt injection. 10 vendors have validated our discoveries, including Notion, which has the potential to impact millions of users. Our investigation illuminates both the possible risks of prompt injection attacks and the possible tactics for mitigation.

Large Language Models (LLMs), like ChatGPT, have demonstrated vast potential but also introduce challenges related to content constraints and potential misuse. Our study investigates three key research questions: (1) the number of different prompt types that can jailbreak LLMs, (2) the effectiveness of jailbreak prompts in circumventing LLM constraints, and (3) the resilience of ChatGPT against these jailbreak prompts. Initially, we develop a classification model to analyze the distribution of existing prompts, identifying ten distinct patterns and three categories of jailbreak prompts. Subsequently, we assess the jailbreak capability of prompts with ChatGPT versions 3.5 and 4.0, utilizing a dataset of 3,120 jailbreak questions across eight prohibited scenarios. Finally, we evaluate the resistance of ChatGPT against jailbreak prompts, finding that the prompts can consistently evade the restrictions in 40 use-case scenarios. The study underscores the importance of prompt structures in jailbreaking LLMs and discusses the challenges of robust jailbreak prompt generation and prevention.

To support software developers in understanding and maintaining programs, various automatic code summarization techniques have been proposed to generate a concise natural language comment for a given code snippet. Recently, the emergence of large language models (LLMs) has led to a great boost in the performance of natural language processing tasks. Among them, ChatGPT is the most popular one which has attracted wide attention from the software engineering community. However, it still remains unclear how ChatGPT performs in (automatic) code summarization. Therefore, in this paper, we focus on evaluating ChatGPT on a widely-used Python dataset called CSN-Python and comparing it with several state-of-the-art (SOTA) code summarization models. Specifically, we first explore an appropriate prompt to guide ChatGPT to generate in-distribution comments. Then, we use such a prompt to ask ChatGPT to generate comments for all code snippets in the CSN-Python test set. We adopt three widely-used metrics (including BLEU, METEOR, and ROUGE-L) to measure the quality of the comments generated by ChatGPT and SOTA models (including NCS, CodeBERT, and CodeT5). The experimental results show that in terms of BLEU and ROUGE-L, ChatGPT's code summarization performance is significantly worse than all three SOTA models. We also present some cases and discuss the advantages and disadvantages of ChatGPT in code summarization. Based on the findings, we outline several open challenges and opportunities in ChatGPT-based code summarization.

In this paper, we present VerifyML, the first secure inference framework to check the fairness degree of a given Machine learning (ML) model. VerifyML is generic and is immune to any obstruction by the malicious model holder during the verification process. We rely on secure two-party computation (2PC) technology to implement VerifyML, and carefully customize a series of optimization methods to boost its performance for both linear and nonlinear layer execution. Specifically, (1) VerifyML allows the vast majority of the overhead to be performed offline, thus meeting the low latency requirements for online inference. (2) To speed up offline preparation, we first design novel homomorphic parallel computing techniques to accelerate the authenticated Beaver's triple (including matrix-vector and convolution triples) generation procedure. It achieves up to $1.7\times$ computation speedup and gains at least $10.7\times$ less communication overhead compared to state-of-the-art work. (3) We also present a new cryptographic protocol to evaluate the activation functions of non-linear layers, which is $4\times$--$42\times$ faster and has $>48\times$ lesser communication than existing 2PC protocol against malicious parties. In fact, VerifyML even beats the state-of-the-art semi-honest ML secure inference system! We provide formal theoretical analysis for VerifyML security and demonstrate its performance superiority on mainstream ML models including ResNet-18 and LeNet.

Dermatological diseases pose a major threat to the global health, affecting almost one-third of the world's population. Various studies have demonstrated that early diagnosis and intervention are often critical to prognosis and outcome. To this end, the past decade has witnessed the rapid evolvement of deep learning based smartphone apps, which allow users to conveniently and timely identify issues that have emerged around their skins. In order to collect sufficient data needed by deep learning and at the same time protect patient privacy, federated learning is often used, where individual clients aggregate a global model while keeping datasets local. However, existing federated learning frameworks are mostly designed to optimize the overall performance, while common dermatological datasets are heavily imbalanced. When applying federated learning to such datasets, significant disparities in diagnosis accuracy may occur. To address such a fairness issue, this paper proposes a fairness-aware federated learning framework for dermatological disease diagnosis. The framework is divided into two stages: In the first in-FL stage, clients with different skin types are trained in a federated learning process to construct a global model for all skin types. An automatic weight aggregator is used in this process to assign higher weights to the client with higher loss, and the intensity of the aggregator is determined by the level of difference between losses. In the latter post-FL stage, each client fine-tune its personalized model based on the global model in the in-FL stage. To achieve better fairness, models from different epochs are selected for each client to keep the accuracy difference of different skin types within 0.05. Experiments indicate that our proposed framework effectively improves both fairness and accuracy compared with the state-of-the-art., Comment: 8 pages, 2 figures

Robotic Vehicles (RVs) have gained great popularity over the past few years. Meanwhile, they are also demonstrated to be vulnerable to sensor spoofing attacks. Although a wealth of research works have presented various attacks, some key questions remain unanswered: are these existing works complete enough to cover all the sensor spoofing threats? If not, how many attacks are not explored, and how difficult is it to realize them? This paper answers the above questions by comprehensively systematizing the knowledge of sensor spoofing attacks against RVs. Our contributions are threefold. (1) We identify seven common attack paths in an RV system pipeline. We categorize and assess existing spoofing attacks from the perspectives of spoofer property, operation, victim characteristic and attack goal. Based on this systematization, we identify 4 interesting insights about spoofing attack designs. (2) We propose a novel action flow model to systematically describe robotic function executions and unexplored sensor spoofing threats. With this model, we successfully discover 103 spoofing attack vectors, 26 of which have been verified by prior works, while 77 attacks are never considered. (3) We design two novel attack methodologies to verify the feasibility of newly discovered spoofing attack vectors.

RESTful APIs are arguably the most popular endpoints for accessing Web services. Blackbox testing is one of the emerging techniques for ensuring the reliability of RESTful APIs. The major challenge in testing RESTful APIs is the need for correct sequences of API operation calls for in-depth testing. To build meaningful operation call sequences, researchers have proposed techniques to learn and utilize the API dependencies based on OpenAPI specifications. However, these techniques either lack the overall awareness of how all the APIs are connected or the flexibility of adaptively fixing the learned knowledge. In this paper, we propose Morest, a model-based RESTful API testing technique that builds and maintains a dynamically updating RESTful-service Property Graph (RPG) to model the behaviors of RESTful-services and guide the call sequence generation. We empirically evaluated Morest and the results demonstrate that Morest can successfully request an average of 152.66%-232.45% more API operations, cover 26.16%-103.24% more lines of code, and detect 40.64%-215.94% more bugs than state-of-the-art techniques. In total, we applied Morest to 6 real-world projects and found 44 bugs (13 of them cannot be detected by existing approaches). Specifically, 2 of the confirmed bugs are from Bitbucket, a famous code management service with more than 6 million users.