Search

Safety concerns of Multimodal large language models (MLLMs) have gradually become an important problem in various applications. Surprisingly, previous works indicate a counter-intuitive phenomenon that using textual unlearning to align MLLMs achieves comparable safety performances with MLLMs trained with image-text pairs. To explain such a counter-intuitive phenomenon, we discover a visual safety information leakage (VSIL) problem in existing multimodal safety benchmarks, i.e., the potentially risky and sensitive content in the image has been revealed in the textual query. In this way, MLLMs can easily refuse these sensitive text-image queries according to textual queries. However, image-text pairs without VSIL are common in real-world scenarios and are overlooked by existing multimodal safety benchmarks. To this end, we construct multimodal visual leakless safety benchmark (VLSBench) preventing visual safety leakage from image to textual query with 2.4k image-text pairs. Experimental results indicate that VLSBench poses a significant challenge to both open-source and close-source MLLMs, including LLaVA, Qwen2-VL, Llama3.2-Vision, and GPT-4o. This study demonstrates that textual alignment is enough for multimodal safety scenarios with VSIL, while multimodal alignment is a more promising solution for multimodal safety scenarios without VSIL. Please see our code and data at: http://hxhcreate.github.io/VLSBench

3D Gaussian Splatting (3DGS) has recently created impressive assets for various applications. However, the copyright of these assets is not well protected as existing watermarking methods are not suited for 3DGS considering security, capacity, and invisibility. Besides, these methods often require hours or even days for optimization, limiting the application scenarios. In this paper, we propose GuardSplat, an innovative and efficient framework that effectively protects the copyright of 3DGS assets. Specifically, 1) We first propose a CLIP-guided Message Decoupling Optimization module for training the message decoder, leveraging CLIP's aligning capability and rich representations to achieve a high extraction accuracy with minimal optimization costs, presenting exceptional capability and efficiency. 2) Then, we propose a Spherical-harmonic-aware (SH-aware) Message Embedding module tailored for 3DGS, which employs a set of SH offsets to seamlessly embed the message into the SH features of each 3D Gaussian while maintaining the original 3D structure. It enables the 3DGS assets to be watermarked with minimal fidelity trade-offs and prevents malicious users from removing the messages from the model files, meeting the demands for invisibility and security. 3) We further propose an Anti-distortion Message Extraction module to improve robustness against various visual distortions. Extensive experiments demonstrate that GuardSplat outperforms the state-of-the-art methods and achieves fast optimization speed., Comment: Project page: https://narcissusex.github.io/GuardSplat and Code: https://github.com/NarcissusEx/GuardSplat

Large Language Models (LLMs) are increasingly used in a variety of applications, but concerns around membership inference have grown in parallel. Previous efforts focus on black-to-grey-box models, thus neglecting the potential benefit from internal LLM information. To address this, we propose the use of Linear Probes (LPs) as a method to detect Membership Inference Attacks (MIAs) by examining internal activations of LLMs. Our approach, dubbed LUMIA, applies LPs layer-by-layer to get fine-grained data on the model inner workings. We test this method across several model architectures, sizes and datasets, including unimodal and multimodal tasks. In unimodal MIA, LUMIA achieves an average gain of 15.71 % in Area Under the Curve (AUC) over previous techniques. Remarkably, LUMIA reaches AUC>60% in 65.33% of cases -- an increment of 46.80% against the state of the art. Furthermore, our approach reveals key insights, such as the model layers where MIAs are most detectable. In multimodal models, LPs indicate that visual inputs can significantly contribute to detect MIAs -- AUC>60% is reached in 85.90% of experiments.

Voice authentication on IoT-enabled smart devices has gained prominence in recent years due to increasing concerns over user privacy and security. The current authentication systems are vulnerable to different voice-spoofing attacks (e.g., replay, voice cloning, and audio deepfakes) that mimic legitimate voices to deceive authentication systems and enable fraudulent activities (e.g., impersonation, unauthorized access, financial fraud, etc.). Existing solutions are often designed to tackle a single type of attack, leading to compromised performance against unseen attacks. On the other hand, existing unified voice anti-spoofing solutions, not designed specifically for IoT, possess complex architectures and thus cannot be deployed on IoT-enabled smart devices. Additionally, most of these unified solutions exhibit significant performance issues, including higher equal error rates or lower accuracy for specific attacks. To overcome these issues, we present the parallel stacked aggregation network (PSA-Net), a lightweight framework designed as an anti-spoofing defense system for voice-controlled smart IoT devices. The PSA-Net processes raw audios directly and eliminates the need for dataset-dependent handcrafted features or pre-computed spectrograms. Furthermore, PSA-Net employs a split-transform-aggregate approach, which involves the segmentation of utterances, the extraction of intrinsic differentiable embeddings through convolutions, and the aggregation of them to distinguish legitimate from spoofed audios. In contrast to existing deep Resnet-oriented solutions, we incorporate cardinality as an additional dimension in our network, which enhances the PSA-Net ability to generalize across diverse attacks. The results show that the PSA-Net achieves more consistent performance for different attacks that exist in current anti-spoofing solutions., Comment: arXiv admin note: text overlap with arXiv:2309.10560

ICS environments are vital to the operation of critical infrastructure such as power grids, water treatment facilities, and manufacturing plants. However, these systems are vulnerable to cyber attacks due to their reliance on interconnected devices and networks, which could lead to catastrophic failures. Therefore, securing these systems from cyber threats becomes paramount. In this context, threat modeling plays an essential role. Despite the advances in threat modeling, the fundamental gap in the state-of-the art is the lack of a systematic methodology for identifying threats in ICS comprehensively. Most threat models in the literature (i) rely on expert knowledge, (ii) only include generic threats such as spoofing, tampering, etc., and (iii) these threats are not comprehensive enough for the systems in question. To overcome these limitations, we propose a novel evidence-based methodology to systematically identify threats based on existing CVE entries of components and their associated fundamental weaknesses in the form of CWE entries - namely, CVE-CWE pairs - and thereby generate a comprehensive threat list. Furthermore, we have implemented our methodology as a ready-to-use tool and have applied it to a typical SCADA system to demonstrate that our methodology is practical and applicable in real-world settings.

In an era marked by the widespread sharing of digital content, the need for a robust content-integrity verification goes beyond the confines of individual social media platforms. While verified profiles (such as blue ticks on platforms like Instagram and X) have become synonymous with credibility, the content they share often traverses a complex network of interconnected platforms, by means of re-sharing, re-posting, etc., leaving a void in the authentication process of the content itself. With the advent of easily accessible AI tools (like DALL-E, Sora, and the tools that are explicitly built for generating deepfakes & face swaps), the risk of misinformation through social media platforms is growing exponentially. This paper discusses a solution, a Content Verification System, designed to authenticate images and videos shared as posts or stories across the digital landscape. Going beyond the limitations of blue ticks, this system empowers individuals and influencers to validate the authenticity of their digital footprint, safeguarding their reputation in an interconnected world.

We describe the Forensics Adapter, an adapter network designed to transform CLIP into an effective and generalizable face forgery detector. Although CLIP is highly versatile, adapting it for face forgery detection is non-trivial as forgery-related knowledge is entangled with a wide range of unrelated knowledge. Existing methods treat CLIP merely as a feature extractor, lacking task-specific adaptation, which limits their effectiveness. To address this, we introduce an adapter to learn face forgery traces -- the blending boundaries unique to forged faces, guided by task-specific objectives. Then we enhance the CLIP visual tokens with a dedicated interaction strategy that communicates knowledge across CLIP and the adapter. Since the adapter is alongside CLIP, its versatility is highly retained, naturally ensuring strong generalizability in face forgery detection. With only $\bm{5.7M}$ trainable parameters, our method achieves a significant performance boost, improving by approximately $\bm{7\%}$ on average across five standard datasets. We believe the proposed method can serve as a baseline for future CLIP-based face forgery detection methods.

Recent advancements in fine-tuning proprietary language models enable customized applications across various domains but also introduce two major challenges: high resource demands and security risks. Regarding resource demands, recent work proposes novel partial compression, such as BitDelta, to quantize the delta weights between the fine-tuned model and base model. Regarding the security risks, user-defined fine-tuning can introduce security vulnerabilities, such as alignment issues, backdoor attacks, and hallucinations. However, most of the current efforts in security assessment focus on the full-precision or full-compression models, it is not well-discussed how the partial compression methods affect security concerns. To bridge this gap, we evaluate the robustness of delta-weight quantization against these security threats. In this paper, we uncover a "free lunch" phenomenon: partial compression can enhance model security against fine-tuning-based attacks with bearable utility loss. Using Llama-2-7b-chat as a case study, we show that, with under 10% utility degradation, the partial compression mitigates alignment-breaking risks by up to 66.17%, harmful backdoor vulnerabilities by 64.46%, and targeted output manipulation risks by up to 90.53%. We further apply LogitLens to visualize internal state transformations during forward passes, suggesting mechanisms for both security failure and recovery in standard versus compressed fine-tuning. This work offers new insights into selecting effective delta compression methods for secure, resource-efficient multi-tenant services.

The advent of instruction-tuned Large Language Models designed for coding tasks (Code LLMs) has transformed software engineering practices. However, their robustness against various input challenges remains a critical concern. This study introduces DegradePrompter, a novel method designed to systematically evaluate the robustness of instruction-tuned Code LLMs. We assess the impact of diverse input challenges on the functionality and correctness of generated code using rigorous metrics and established benchmarks. Our comprehensive evaluation includes five state-of-the-art open-source models and three production-grade closed-source models, revealing varying degrees of robustness. Open-source models demonstrate an increased susceptibility to input perturbations, resulting in declines in functional correctness ranging from 12% to 34%. In contrast, commercial models demonstrate relatively greater resilience, with performance degradation ranging from 3% to 24%. To enhance the robustness of the models against these vulnerabilities, we investigate a straightforward yet effective mitigation strategy. Our findings highlight the need for robust defense mechanisms and comprehensive evaluations during both the development and deployment phases to ensure the resilience and reliability of automated code generation systems.

Deep neural networks (DNNs) are susceptible to backdoor attacks, where adversaries poison datasets with adversary-specified triggers to implant hidden backdoors, enabling malicious manipulation of model predictions. Dataset purification serves as a proactive defense by removing malicious training samples to prevent backdoor injection at its source. We first reveal that the current advanced purification methods rely on a latent assumption that the backdoor connections between triggers and target labels in backdoor attacks are simpler to learn than the benign features. We demonstrate that this assumption, however, does not always hold, especially in all-to-all (A2A) and untargeted (UT) attacks. As a result, purification methods that analyze the separation between the poisoned and benign samples in the input-output space or the final hidden layer space are less effective. We observe that this separability is not confined to a single layer but varies across different hidden layers. Motivated by this understanding, we propose FLARE, a universal purification method to counter various backdoor attacks. FLARE aggregates abnormal activations from all hidden layers to construct representations for clustering. To enhance separation, FLARE develops an adaptive subspace selection algorithm to isolate the optimal space for dividing an entire dataset into two clusters. FLARE assesses the stability of each cluster and identifies the cluster with higher stability as poisoned. Extensive evaluations on benchmark datasets demonstrate the effectiveness of FLARE against 22 representative backdoor attacks, including all-to-one (A2O), all-to-all (A2A), and untargeted (UT) attacks, and its robustness to adaptive attacks., Comment: 13 pages

Blockchains have sparked global interest in recent years, gaining importance as they increasingly influence technology and finance. This thesis investigates the robustness of blockchain protocols, specifically focusing on Ethereum Proof-of-Stake. We define robustness in terms of two critical properties: Safety, which ensures that the blockchain will not have permanent conflicting blocks, and Liveness, which guarantees the continuous addition of new reliable blocks. Our research addresses the gap between traditional distributed systems approaches, which classify agents as either honest or Byzantine (i.e., malicious or faulty), and game-theoretic models that consider rational agents driven by incentives. We explore how incentives impact the robustness with both approaches. The thesis comprises three distinct analyses. First, we formalize the Ethereum PoS protocol, defining its properties and examining potential vulnerabilities through a distributed systems perspective. We identify that certain attacks can undermine the system's robustness. Second, we analyze the inactivity leak mechanism, a critical feature of Ethereum PoS, highlighting its role in maintaining system liveness during network disruptions but at the cost of safety. Finally, we employ game-theoretic models to study the strategies of rational validators within Ethereum PoS, identifying conditions under which these agents might deviate from the prescribed protocol to maximize their rewards. Our findings contribute to a deeper understanding of the importance of incentive mechanisms for blockchain robustness and provide insights into designing more resilient blockchain protocols., Comment: PhD thesis

The evolving threat landscape in cybersecurity necessitates the adoption of advanced tools for effective vulnerability management. This paper presents a comprehensive comparative analysis of three widely used tools: Nessus, Acunetix, and Nikto. Each tool is assessed based on its detection accuracy, risk scoring using the Common Vulnerability Scoring System (CVSS), ease of use, automation and reporting capabilities, performance metrics, and cost effectiveness. The research addresses the challenges faced by organizations in selecting the most suitable tool for their unique security requirements.

Current black-box backdoor attacks in convolutional neural networks formulate attack objective(s) as single-objective optimization problems in single domain. Designing triggers in single domain harms semantics and trigger robustness as well as introduces visual and spectral anomaly. This work proposes a multi-objective black-box backdoor attack in dual domains via evolutionary algorithm (LADDER), the first instance of achieving multiple attack objectives simultaneously by optimizing triggers without requiring prior knowledge about victim model. In particular, we formulate LADDER as a multi-objective optimization problem (MOP) and solve it via multi-objective evolutionary algorithm (MOEA). MOEA maintains a population of triggers with trade-offs among attack objectives and uses non-dominated sort to drive triggers toward optimal solutions. We further apply preference-based selection to MOEA to exclude impractical triggers. We state that LADDER investigates a new dual-domain perspective for trigger stealthiness by minimizing the anomaly between clean and poisoned samples in the spectral domain. Lastly, the robustness against preprocessing operations is achieved by pushing triggers to low-frequency regions. Extensive experiments comprehensively showcase that LADDER achieves attack effectiveness of at least 99%, attack robustness with 90.23% (50.09% higher than state-of-the-art attacks on average), superior natural stealthiness (1.12x to 196.74x improvement) and excellent spectral stealthiness (8.45x enhancement) as compared to current stealthy attacks by the average $l_2$-norm across 5 public datasets.

Nowadays, inter-vehicle networks are a viable communication scenario that greatly contributes to daily work, and its issues are gaining more and more attention every day. These days, space networks are growing and developing. There are numerous new uses for this new kind of network communication. One of the most significant daily programs in the world today is road traffic. For human growth, passenger and freight transportation is essential. Thus, fresh advancements in the areas of improved safety features, environmentally friendly fuel, etc., are developed daily. In order to improve safety and regulate traffic, a new application program is used. However, because of their stringent security standards, these initiatives have an impact on traffic safety. Since driving is one of the things that necessitates traffic safety, this area needs to be made more secure. Providing trustworthy driving data is crucial to achieving this goal, aside from the automated portion of the operation. Drivers would greatly benefit from accurate weather descriptions or early warnings of potential dangers (such as traffic bottlenecks or accidents). Inter-vehicle networks, a novel form of information technology, are being developed for this reason. Keywords: inter-vehicle network, transportation and security, Comment: 7 pages, 3 figures

As Large Language Models (LLMs) are progressively deployed across diverse fields and real-world applications, ensuring the security and robustness of LLMs has become ever more critical. Retrieval-Augmented Generation (RAG) is a cutting-edge approach designed to address the limitations of large language models (LLMs). By retrieving information from the relevant knowledge database, RAG enriches the input to LLMs, enabling them to produce responses that are more accurate and contextually appropriate. It is worth noting that the knowledge database, being sourced from publicly available channels such as Wikipedia, inevitably introduces a new attack surface. RAG poisoning involves injecting malicious texts into the knowledge database, ultimately leading to the generation of the attacker's target response (also called poisoned response). However, there are currently limited methods available for detecting such poisoning attacks. We aim to bridge the gap in this work. Particularly, we introduce RevPRAG, a flexible and automated detection pipeline that leverages the activations of LLMs for poisoned response detection. Our investigation uncovers distinct patterns in LLMs' activations when generating correct responses versus poisoned responses. Our results on multiple benchmark datasets and RAG architectures show our approach could achieve 98% true positive rate, while maintaining false positive rates close to 1%. We also evaluate recent backdoor detection methods specifically designed for LLMs and applicable for identifying poisoned responses in RAG. The results demonstrate that our approach significantly surpasses them.

The decentralized exchange (DEX) leverages smart contracts to trade digital assets for users on the blockchain. Developers usually develop several smart contracts into one project, implementing complex logic functions and multiple transaction operations. However, the interaction among these contracts poses challenges for developers analyzing the state logic. Due to the complex state logic in DEX projects, many critical state derailment defects have emerged in recent years. In this paper, we conduct the first systematic study of state derailment defects in DEX. We define five categories of state derailment defects and provide detailed analyses of them. Furthermore, we propose a novel deep learning-based framework StateGuard for detecting state derailment defects in DEX smart contracts. It leverages a smart contract deconstructor to deconstruct the contract into an Abstract Syntax Tree (AST), from which five categories of dependency features are extracted. Next, it implements a graph optimizer to process the structured data. At last, the optimized data is analyzed by Graph Convolutional Networks (GCNs) to identify potential state derailment defects. We evaluated StateGuard through a dataset of 46 DEX projects containing 5,671 smart contracts, and it achieved 94.25% F1-score. In addition, in a comparison experiment with state-of-the-art, StateGuard leads the F1-score by 6.29%. To further verify its practicality, we used StateGuar to audit real-world contracts and successfully authenticated multiple novel CVEs., Comment: 13 pages

JavaScript has been consistently among the most popular programming languages in the past decade. However, its dynamic, weakly-typed, and asynchronous nature can make it challenging to write maintainable code for developers without in-depth knowledge of the language. Consequently, many JavaScript applications tend to contain code smells that adversely influence program comprehension, maintenance, and debugging. Due to the widespread usage of JavaScript, code security is an important matter. While JavaScript code smells and detection techniques have been studied in the past, current work on security smells for JavaScript is scarce. Security code smells are coding patterns indicative of potential vulnerabilities or security weaknesses. Identifying security code smells can help developers to focus on areas where additional security measures may be needed. We present a set of 24 JavaScript security code smells, map them to a possible security awareness defined by Common Weakness Enumeration (CWE), explain possible refactoring, and explain our detection mechanism. We implement our security code smell detection on top of an existing open source tool that was proposed to detect general code smells in JavaScript., Comment: 9 pages

Dynamic taint tracking is the process of assigning label to variables in a program and then tracking the flow of the labels as the program executes. Dynamic taint tracking for java applications is achieved by instrumenting the application ie. adding parallel variable for each actual variable of the program and inserting additional bytecode instructions to track the flow of the parallel variables. In this paper we suggest partial instrumentation to achieve dynamic taint tracking with reasonable runtime overhead. Partial instrumentation involves instrumenting only parts of a java application, which are within the scope of a predefined source and sink set. Partial instrumentation is performed at the granularity level of a method. We use PetaBlox, a large-scale software analysis tool, which internally uses Datalog[3], to perform static analysis and infers all the methods within the scope of source and sink sets and a modified version of Phosphor[1] to achieve partial instrumentation. Test runs performed on some of the Dacapo benchmarks show a significant performance improvement over the version of Phosphor that performs complete instrumentation.

This article presents examples of an application of the finite field method for the computation of the characteristic polynomial of the matching arrangement of a graph. Weight functions on edges of a graph with weights from a finite field are divided into proper and improper functions in connection with proper colorings of vertices of the matching polytope of a graph. An improper weight function problem is introduced, a proof of its NP-completeness is presented, and a knapsack-like public key cryptosystem is constructed based on the improper weight function problem.

Federated Parameter-Efficient Fine-Tuning (FedPEFT) has emerged as a promising paradigm for privacy-preserving and efficient adaptation of Pre-trained Language Models (PLMs) in Federated Learning (FL) settings. It preserves data privacy by keeping the data decentralized and training the model on local devices, ensuring that raw data never leaves the user's device. Moreover, the integration of PEFT methods such as LoRA significantly reduces the number of trainable parameters compared to fine-tuning the entire model, thereby minimizing communication costs and computational overhead. Despite its potential, the security implications of FedPEFT remain underexplored. This paper introduces a novel security threat to FedPEFT, termed PEFT-as-an-Attack (PaaA), which exposes how PEFT can be exploited as an attack vector to circumvent PLMs' safety alignment and generate harmful content in response to malicious prompts. Our evaluation of PaaA reveals that with less than 1% of the model's parameters set as trainable, and a small subset of clients acting maliciously, the attack achieves an approximate 80% attack success rate using representative PEFT methods such as LoRA. To mitigate this threat, we further investigate potential defense strategies, including Robust Aggregation Schemes (RASs) and Post-PEFT Safety Alignment (PPSA). However, our empirical analysis highlights the limitations of these defenses, i.e., even the most advanced RASs, such as DnC and ClippedClustering, struggle to defend against PaaA in scenarios with highly heterogeneous data distributions. Similarly, while PPSA can reduce attack success rates to below 10%, it severely degrades the model's accuracy on the target task. Our results underscore the urgent need for more effective defense mechanisms that simultaneously ensure security and maintain the performance of the FedPEFT paradigm.

Smart contracts are essential for managing digital assets in blockchain networks, highlighting the need for effective security measures. This paper introduces SmartLLMSentry, a novel framework that leverages large language models (LLMs), specifically ChatGPT with in-context training, to advance smart contract vulnerability detection. Traditional rule-based frameworks have limitations in integrating new detection rules efficiently. In contrast, SmartLLMSentry utilizes LLMs to streamline this process. We created a specialized dataset of five randomly selected vulnerabilities for model training and evaluation. Our results show an exact match accuracy of 91.1% with sufficient data, although GPT-4 demonstrated reduced performance compared to GPT-3 in rule generation. This study illustrates that SmartLLMSentry significantly enhances the speed and accuracy of vulnerability detection through LLMdriven rule integration, offering a new approach to improving Blockchain security and addressing previously underexplored vulnerabilities in smart contracts.

Large Language Models (LLMs) are increasingly used in software development to generate functions, such as attack detectors, that implement security requirements. However, LLMs struggle to generate accurate code, resulting, e.g., in attack detectors that miss well-known attacks when used in practice. This is most likely due to the LLM lacking knowledge about some existing attacks and to the generated code being not evaluated in real usage scenarios. We propose a novel approach integrating Retrieval Augmented Generation (RAG) and Self-Ranking into the LLM pipeline. RAG enhances the robustness of the output by incorporating external knowledge sources, while the Self-Ranking technique, inspired to the concept of Self-Consistency, generates multiple reasoning paths and creates ranks to select the most robust detector. Our extensive empirical study targets code generated by LLMs to detect two prevalent injection attacks in web security: Cross-Site Scripting (XSS) and SQL injection (SQLi). Results show a significant improvement in detection performance compared to baselines, with an increase of up to 71%pt and 37%pt in the F2-Score for XSS and SQLi detection, respectively.

Large language models (LLMs) possess extensive knowledge and question-answering capabilities, having been widely deployed in privacy-sensitive domains like finance and medical consultation. During LLM inferences, cache-sharing methods are commonly employed to enhance efficiency by reusing cached states or responses for the same or similar inference requests. However, we identify that these cache mechanisms pose a risk of private input leakage, as the caching can result in observable variations in response times, making them a strong candidate for a timing-based attack hint. In this study, we propose a novel timing-based side-channel attack to execute input theft in LLMs inference. The cache-based attack faces the challenge of constructing candidate inputs in a large search space to hit and steal cached user queries. To address these challenges, we propose two primary components. The input constructor employs machine learning techniques and LLM-based approaches for vocabulary correlation learning while implementing optimized search mechanisms for generalized input construction. The time analyzer implements statistical time fitting with outlier elimination to identify cache hit patterns, continuously providing feedback to refine the constructor's search strategy. We conduct experiments across two cache mechanisms and the results demonstrate that our approach consistently attains high attack success rates in various applications. Our work highlights the security vulnerabilities associated with performance optimizations, underscoring the necessity of prioritizing privacy and security alongside enhancements in LLM inference.

Greybox fuzzing has emerged as a preferred technique for discovering software bugs, striking a balance between efficiency and depth of exploration. While research has focused on improving fuzzing techniques, the importance of high-quality initial seeds remains critical yet often overlooked. Existing methods for seed generation are limited, especially for programs with non-standard or custom input formats. Large Language Models (LLMs) has revolutionized numerous domains, showcasing unprecedented capabilities in understanding and generating complex patterns across various fields of knowledge. This paper introduces SeedMind, a novel system that leverages LLMs to boost greybox fuzzing through intelligent seed generation. Unlike previous approaches, SeedMind employs LLMs to create test case generators rather than directly producing test cases. Our approach implements an iterative, feedback-driven process that guides the LLM to progressively refine test case generation, aiming for increased code coverage depth and breadth. In developing SeedMind, we addressed key challenges including input format limitations, context window constraints, and ensuring consistent, progress-aware behavior. Intensive evaluations with real-world applications show that SeedMind effectively harnesses LLMs to generate high-quality test cases and facilitate fuzzing in bug finding, presenting utility comparable to human-created seeds and significantly outperforming the existing LLM-based solutions.

The efficacy of address space layout randomization has been formally demonstrated in a shared-memory model by Abadi et al., contingent on specific assumptions about victim programs. However, modern operating systems, implementing layout randomization in the kernel, diverge from these assumptions and operate on a separate memory model with communication through system calls. In this work, we relax Abadi et al.'s language assumptions while demonstrating that layout randomization offers a comparable safety guarantee in a system with memory separation. However, in practice, speculative execution and side-channels are recognized threats to layout randomization. We show that kernel safety cannot be restored for attackers capable of using side-channels and speculative execution, and introduce enforcement mechanisms that can guarantee speculative kernel safety for safe system calls in the Spectre era. We implement two suitable mechanisms and we use them to compile the Linux kernel in order to evaluate their performance overhead., Comment: arXiv admin note: substantial text overlap with arXiv:2406.07278

In 2022, a prominent supersingular isogeny-based cryptographic scheme, namely SIDH, was compromised by a key recovery attack. However, this attack does not undermine the isogeny path problem, which remains central to the security of isogeny-based cryptography. Following the attacks by Castryck and Decru, as well as Maino and Martindale, Robert gave a mature and polynomial-time algorithm that transforms the SIDH key recovery attack into a valuable cryptographic tool. In this paper, we combine this tool with advanced encoding techniques to construct a novel threshold scheme.

Verifying the correctness of a digital twin provides a formal guarantee that the digital twin operates as intended. Digital twin verification is challenging due to the presence of uncertainties in the virtual representation, the physical environment, and the bidirectional flow of information between physical and virtual. A further challenge is that a digital twin of a complex system is composed of distributed components. This paper presents a methodology to specify and verify digital twin behavior, translating uncertain processes into a formally verifiable finite state machine. We use the Temporal Logic of Actions (TLA) to create a specification, an implementation abstraction that defines the properties required for correct system behavior. Our approach includes a novel weakening of formal security properties, allowing controlled information leakage while preserving theoretical guarantees. We demonstrate this approach on a digital twin of an unmanned aerial vehicle, verifying synchronization of physical-to-virtual and virtual-to-digital data flows to detect unintended misalignments., Comment: 23 pages

Adversarial input image perturbation attacks have emerged as a significant threat to machine learning algorithms, particularly in image classification setting. These attacks involve subtle perturbations to input images that cause neural networks to misclassify the input images, even though the images remain easily recognizable to humans. One critical area where adversarial attacks have been demonstrated is in automotive systems where traffic sign classification and recognition is critical, and where misclassified images can cause autonomous systems to take wrong actions. This work presents a new class of adversarial attacks. Unlike existing work that has focused on adversarial perturbations that leverage human-made artifacts to cause the perturbations, such as adding stickers, paint, or shining flashlights at traffic signs, this work leverages nature-made artifacts: tree leaves. By leveraging nature-made artifacts, the new class of attacks has plausible deniability: a fall leaf stuck to a street sign could come from a near-by tree, rather than be placed there by an malicious human attacker. To evaluate the new class of the adversarial input image perturbation attacks, this work analyses how fall leaves can cause misclassification in street signs. The work evaluates various leaves from different species of trees, and considers various parameters such as size, color due to tree leaf type, and rotation. The work demonstrates high success rate for misclassification. The work also explores the correlation between successful attacks and how they affect the edge detection, which is critical in many image classification algorithms.

Understanding the attack patterns associated with a cyberattack is crucial for comprehending the attacker's behaviors and implementing the right mitigation measures. However, majority of the information regarding new attacks is typically presented in unstructured text, posing significant challenges for security analysts in collecting necessary information. In this paper, we present a sentence classification system that can identify the attack techniques described in natural language sentences from cyber threat intelligence (CTI) reports. We propose a new method for utilizing auxiliary data with the same labels to improve classification for the low-resource cyberattack classification task. The system first trains the model using the augmented training data and then trains more using only the primary data. We validate our model using the TRAM data1 and the MITRE ATT&CK framework. Experiments show that our method enhances Macro-F1 by 5 to 9 percentage points and keeps Micro-F1 scores competitive when compared to the baseline performance on the TRAM dataset.

Ensuring privacy during inference stage is crucial to prevent malicious third parties from reconstructing users' private inputs from outputs of public models. Despite a large body of literature on privacy preserving learning (which ensures privacy of training data), there is no existing systematic framework to ensure the privacy of users' data during inference. Motivated by this problem, we introduce the notion of Inference Privacy (IP), which can allow a user to interact with a model (for instance, a classifier, or an AI-assisted chat-bot) while providing a rigorous privacy guarantee for the users' data at inference. We establish fundamental properties of the IP privacy notion and also contrast it with the notion of Local Differential Privacy (LDP). We then present two types of mechanisms for achieving IP: namely, input perturbations and output perturbations which are customizable by the users and can allow them to navigate the trade-off between utility and privacy. We also demonstrate the usefulness of our framework via experiments and highlight the resulting trade-offs between utility and privacy during inference.

The Single-Turn Crescendo Attack (STCA), first introduced in Aqrawi and Abbasi [2024], is an innovative method designed to bypass the ethical safeguards of text-to-text AI models, compelling them to generate harmful content. This technique leverages a strategic escalation of context within a single prompt, combined with trust-building mechanisms, to subtly deceive the model into producing unintended outputs. Extending the application of STCA to text-to-image models, we demonstrate its efficacy by compromising the guardrails of a widely-used model, DALL-E 3, achieving outputs comparable to outputs from the uncensored model Flux Schnell, which served as a baseline control. This study provides a framework for researchers to rigorously evaluate the robustness of guardrails in text-to-image models and benchmark their resilience against adversarial attacks.

With the widespread deployment of Multimodal Large Language Models (MLLMs) for visual-reasoning tasks, improving their safety has become crucial. Recent research indicates that despite training-time safety alignment, these models remain vulnerable to jailbreak attacks: carefully crafted image-prompt pairs that compel the model to generate harmful content. In this work, we first highlight a critical safety gap, demonstrating that alignment achieved solely through safety training may be insufficient against jailbreak attacks. To address this vulnerability, we propose Immune, an inference-time defense framework that leverages a safe reward model during decoding to defend against jailbreak attacks. Additionally, we provide a rigorous mathematical characterization of Immune, offering provable guarantees against jailbreaks. Extensive evaluations on diverse jailbreak benchmarks using recent MLLMs reveal that Immune effectively enhances model safety while preserving the model's original capabilities. For instance, against text-based jailbreak attacks on LLaVA-1.6, Immune reduces the attack success rate by 57.82% and 16.78% compared to the base MLLM and state-of-the-art defense strategy, respectively.

With the development of the internet, recommending interesting products to users has become a highly valuable research topic for businesses. Recommendation systems play a crucial role in addressing this issue. To prevent the leakage of each user's (client's) private data, Federated Recommendation Systems (FedRec) have been proposed and widely used. However, extensive research has shown that FedRec suffers from security issues such as data privacy leakage, and it is challenging to train effective models with FedRec when each client only holds interaction information for a single user. To address these two problems, this paper proposes a new privacy-preserving recommendation system (PRSI), which includes a preprocessing module and two main phases. The preprocessing module employs split vectors and fake interaction items to protect clients' interaction information and recommendation results. The two main phases are: (1) the collection of interaction information and (2) the sending of recommendation results. In the interaction information collection phase, each client uses the preprocessing module and random communication methods (according to the designed interactive protocol) to protect their ID information and IP addresses. In the recommendation results sending phase, the central server uses the preprocessing module and triplets to distribute recommendation results to each client under secure conditions, following the designed interactive protocol. Finally, we conducted multiple sets of experiments to verify the security, accuracy, and communication cost of the proposed method.

A strategy used by malicious actors is to "live off the land," where benign systems and tools already available on a victim's systems are used and repurposed for the malicious actor's intent. In this work, we ask if there is a way for anti-virus developers to similarly re-purpose existing work to improve their malware detection capability. We show that this is plausible via YARA rules, which use human-written signatures to detect specific malware families, functionalities, or other markers of interest. By extracting sub-signatures from publicly available YARA rules, we assembled a set of features that can more effectively discriminate malicious samples from benign ones. Our experiments demonstrate that these features add value beyond traditional features on the EMBER 2018 dataset. Manual analysis of the added sub-signatures shows a power-law behavior in a combination of features that are specific and unique, as well as features that occur often. A prior expectation may be that the features would be limited in being overly specific to unique malware families. This behavior is observed, and is apparently useful in practice. In addition, we also find sub-signatures that are dual-purpose (e.g., detecting virtual machine environments) or broadly generic (e.g., DLL imports)., Comment: To appear in BigData'24 CyberHunt 2024

As the outputs of generative AI (GenAI) techniques improve in quality, it becomes increasingly challenging to distinguish them from human-created content. Watermarking schemes are a promising approach to address the problem of distinguishing between AI and human-generated content. These schemes embed hidden signals within AI-generated content to enable reliable detection. While watermarking is not a silver bullet for addressing all risks associated with GenAI, it can play a crucial role in enhancing AI safety and trustworthiness by combating misinformation and deception. This paper presents a comprehensive overview of watermarking techniques for GenAI, beginning with the need for watermarking from historical and regulatory perspectives. We formalize the definitions and desired properties of watermarking schemes and examine the key objectives and threat models for existing approaches. Practical evaluation strategies are also explored, providing insights into the development of robust watermarking techniques capable of resisting various attacks. Additionally, we review recent representative works, highlight open challenges, and discuss potential directions for this emerging field. By offering a thorough understanding of watermarking in GenAI, this work aims to guide researchers in advancing watermarking methods and applications, and support policymakers in addressing the broader implications of GenAI.

Client diversity in the Ethereum blockchain refers to the use of multiple independent implementations of the Ethereum protocol. This effectively enhances network resilience by reducing reliance on any single software client implementation. With client diversity, a single bug cannot tear the whole network down. However, despite multiple production-grade client implementations being available, there is still a heavily skewed distribution of clients in Ethereum. This is a concern for the community. In this paper, we introduce a novel conceptual framework for client diversity. The core goal is to improve the network resilience as a systemic property. Our key insight is to leverage economic incentives and verifiable execution to encourage the adoption of minority clients, thereby fostering a more robust blockchain ecosystem. Concretely, we propose to unambiguously and provably identify the client implementation used by any protocol participant, and to use this information to incentivize the usage of minority clients by offering higher participation rewards. We outline a detailed blueprint for our conceptual framework, in the realm of Ethereum. Our proposal is a game changer for improving client diversity of blockchains. Ultimately, it applies to strengthening the resilience of any decentralized distributed systems.

HPC systems face security and compliance challenges, particularly in preventing waste and misuse of computational resources by unauthorized or malicious software that deviates from allocation purpose. Existing methods to classify applications based on job names or resource usage are often unreliable or fail to capture applications that have different behavior due to different inputs or system noise. This research proposes an approach that uses similarity-preserving fuzzy hashes to classify HPC application executables. By comparing the similarity of SSDeep fuzzy hashes, a Random Forest Classifier can accurately label applications executing on HPC systems including unknown samples. We evaluate the Fuzzy Hash Classifier on a dataset of 92 application classes and 5333 distinct application samples. The proposed method achieved a macro f1-score of 90% (micro f1-score: 89%, weighted f1-score: 90%). Our approach addresses the critical need for more effective application classification in HPC environments, minimizing resource waste, and enhancing security and compliance.

Designing resilient Internet of Things (IoT) systems requires i) identification of IoT Critical Objects (ICOs) such as services, devices, and resources, ii) threat analysis, and iii) mitigation strategy selection. However, the traditional process for designing resilient IoT systems is still manual, leading to inefficiencies and increased risks. In addition, while tools such as ChatGPT could support this manual and highly error-prone process, their use raises concerns over data privacy, inconsistent outputs, and internet dependence. Therefore, we propose RITA, an automated, open-source framework that uses a fine-tuned RoBERTa-based Named Entity Recognition (NER) model to identify ICOs from IoT requirement documents, correlate threats, and recommend countermeasures. RITA operates entirely offline and can be deployed on-site, safeguarding sensitive information and delivering consistent outputs that enhance standardization. In our empirical evaluation, RITA outperformed ChatGPT in four of seven ICO categories, particularly in actuator, sensor, network resource, and service identification, using both human-annotated and ChatGPT-generated test data. These findings indicate that RITA can improve resilient IoT design by effectively supporting key security operations, offering a practical solution for developing robust IoT architectures.

Federated Learning (FL) emerged as a paradigm for conducting machine learning across broad and decentralized datasets, promising enhanced privacy by obviating the need for direct data sharing. However, recent studies show that attackers can steal private data through model manipulation or gradient analysis. Existing attacks are constrained by low theft quantity or low-resolution data, and they are often detected through anomaly monitoring in gradients or weights. In this paper, we propose a novel data-reconstruction attack leveraging malicious code injection, supported by two key techniques, i.e., distinctive and sparse encoding design and block partitioning. Unlike conventional methods that require detectable changes to the model, our method stealthily embeds a hidden model using parameter sharing to systematically extract sensitive data. The Fibonacci-based index design ensures efficient, structured retrieval of memorized data, while the block partitioning method enhances our method's capability to handle high-resolution images by dividing them into smaller, manageable units. Extensive experiments on 4 datasets confirmed that our method is superior to the five state-of-the-art data-reconstruction attacks under the five respective detection methods. Our method can handle large-scale and high-resolution data without being detected or mitigated by state-of-the-art data reconstruction defense methods. In contrast to baselines, our method can be directly applied to both FedAVG and FedSGD scenarios, underscoring the need for developers to devise new defenses against such vulnerabilities. We will open-source our code upon acceptance.

Battery-powered embedded systems (BESs) have become ubiquitous. Their internals include a battery management system (BMS), a radio interface, and a motor controller. Despite their associated risk, there is little research on BES internal attack surfaces. To fill this gap, we present the first security and privacy assessment of e-scooters internals. We cover Xiaomi M365 (2016) and ES3 (2023) e-scooters and their interactions with Mi Home (their companion app). We extensively RE their internals and uncover four critical design vulnerabilities, including a remote code execution issue with their BMS. Based on our RE findings, we develop E-Trojans, four novel attacks targeting BES internals. The attacks can be conducted remotely or in wireless proximity. They have a widespread real-world impact as they violate the Xiaomi e-scooter ecosystem safety, security, availability, and privacy. For instance, one attack allows the extortion of money from a victim via a BMS undervoltage battery ransomware. A second one enables user tracking by fingerprinting the BES internals. With extra RE efforts, the attacks can be ported to other BES featuring similar vulnerabilities. We implement our attacks and RE findings in E-Trojans, a modular and low-cost toolkit to test BES internals. Our toolkit binary patches BMS firmware by adding malicious capabilities. It also implements our undervoltage battery ransomware in an Android app with a working backend. We successfully test our four attacks on M365 and ES3, empirically confirming their effectiveness and practicality. We propose four practical countermeasures to fix our attacks and improve the Xiaomi e-scooter ecosystem security and privacy.

Package delivery is a critical aspect of various industries, but it often incurs high financial costs and inefficiencies when relying solely on human resources. The last-mile transport problem, in particular, contributes significantly to the expenditure of human resources in major companies. Robot-based delivery systems have emerged as a potential solution for last-mile delivery to address this challenge. However, robotic delivery systems still face security and privacy issues, like impersonation, replay, man-in-the-middle attacks (MITM), unlinkability, and identity theft. In this context, we propose a privacy-preserving multi-factor authentication scheme specifically designed for robot delivery systems. Additionally, AI-assisted robotic delivery systems are susceptible to machine learning-based attacks (e.g. FGSM, PGD, etc.). We introduce the \emph{first} transformer-based audio-visual fusion defender to tackle this issue, which effectively provides resilience against adversarial samples. Furthermore, we provide a rigorous formal analysis of the proposed protocol and also analyse the protocol security using a popular symbolic proof tool called ProVerif and Scyther. Finally, we present a real-world implementation of the proposed robotic system with the computation cost and energy consumption analysis. Code and pre-trained models are available at: https://drive.google.com/drive/folders/18B2YbxtV0Pyj5RSFX-ZzCGtFOyorBHil

Detecting energy theft is vital for effectively managing power grids, as it ensures precise billing and prevents financial losses. Split-learning emerges as a promising decentralized machine learning technique for identifying energy theft while preserving user data confidentiality. Nevertheless, traditional split learning approaches are vulnerable to privacy leakage attacks, which significantly threaten data confidentiality. To address this challenge, we propose a novel GAN-Transformer-based split learning framework in this paper. This framework leverages the strengths of the transformer architecture, which is known for its capability to process long-range dependencies in energy consumption data. Thus, it enhances the accuracy of energy theft detection without compromising user privacy. A distinctive feature of our approach is the deployment of a novel mask-based method, marking a first in its field to effectively combat privacy leakage in split learning scenarios targeted at AI-enabled adversaries. This method protects sensitive information during the model's training phase. Our experimental evaluations indicate that the proposed framework not only achieves accuracy levels comparable to conventional methods but also significantly enhances privacy protection. The results underscore the potential of the GAN-Transformer split learning framework as an effective and secure tool in the domain of energy theft detection.

We study a classical problem in private prediction, the problem of computing an $(m\epsilon, \delta)$-differentially private majority of $K$ $(\epsilon, \Delta)$-differentially private algorithms for $1 \leq m \leq K$ and $1 > \delta \geq \Delta \geq 0$. Standard methods such as subsampling or randomized response are widely used, but do they provide optimal privacy-utility tradeoffs? To answer this, we introduce the Data-dependent Randomized Response Majority (DaRRM) algorithm. It is parameterized by a data-dependent noise function $\gamma$, and enables efficient utility optimization over the class of all private algorithms, encompassing those standard methods. We show that maximizing the utility of an $(m\epsilon, \delta)$-private majority algorithm can be computed tractably through an optimization problem for any $m \leq K$ by a novel structural result that reduces the infinitely many privacy constraints into a polynomial set. In some settings, we show that DaRRM provably enjoys a privacy gain of a factor of 2 over common baselines, with fixed utility. Lastly, we demonstrate the strong empirical effectiveness of our first-of-its-kind privacy-constrained utility optimization for ensembling labels for private prediction from private teachers in image classification. Notably, our DaRRM framework with an optimized $\gamma$ exhibits substantial utility gains when compared against several baselines., Comment: 57 pages, 10 figures. Proceedings of Transactions on Machine Learning Research (TMLR), November 2024

Adversarial training has emerged as an effective approach to train robust neural network models that are resistant to adversarial attacks, even in low-label regimes where labeled data is scarce. In this paper, we introduce a novel semi-supervised adversarial training approach that enhances both robustness and natural accuracy by generating effective adversarial examples. Our method begins by applying linear interpolation between clean and adversarial examples to create interpolated adversarial examples that cross decision boundaries by a controlled margin. This sample-aware strategy tailors adversarial examples to the characteristics of each data point, enabling the model to learn from the most informative perturbations. Additionally, we propose a global epsilon scheduling strategy that progressively adjusts the upper bound of perturbation strengths during training. The combination of these strategies allows the model to develop increasingly complex decision boundaries with better robustness and natural accuracy. Empirical evaluations show that our approach effectively enhances performance against various adversarial attacks, such as PGD and AutoAttack.

Deep Neural Networks exhibit inherent vulnerabilities to adversarial attacks, which can significantly compromise their outputs and reliability. While existing research primarily focuses on attacking single-task scenarios or indiscriminately targeting all tasks in multi-task environments, we investigate selectively targeting one task while preserving performance in others within a multi-task framework. This approach is motivated by varying security priorities among tasks in real-world applications, such as autonomous driving, where misinterpreting critical objects (e.g., signs, traffic lights) poses a greater security risk than minor depth miscalculations. Consequently, attackers may hope to target security-sensitive tasks while avoiding non-critical tasks from being compromised, thus evading being detected before compromising crucial functions. In this paper, we propose a method for the stealthy multi-task attack framework that utilizes multiple algorithms to inject imperceptible noise into the input. This novel method demonstrates remarkable efficacy in compromising the target task while simultaneously maintaining or even enhancing performance across non-targeted tasks - a criterion hitherto unexplored in the field. Additionally, we introduce an automated approach for searching the weighting factors in the loss function, further enhancing attack efficiency. Experimental results validate our framework's ability to successfully attack the target task while preserving the performance of non-targeted tasks. The automated loss function weight searching method demonstrates comparable efficacy to manual tuning, establishing a state-of-the-art multi-task attack framework.

While the Web has become a worldwide platform for communication, hackers and hacktivists share their ideology and communicate with members on the "Dark Web" - the reverse of the Web. Currently, the problems of information overload and difficulty to obtain a comprehensive picture of hackers and cyber-attackers hinder the effective analysis of predicting their activities on the Web. Also, there are currently more objects connected to the internet than there are people in the world and this gap will continue to grow as more and more objects gain ability to directly interface with the Internet. Many technical communities are vigorously pursuing research topics that contribute to the Internet of Things (IoT). In this paper we have proposed a novel methodology for collecting and analyzing the Dark Web information to identify websites of hackers from the Web sea, and how this information can help us in predicting IoT vulnerabilities. This methodology incorporates information collection, analysis, visualization techniques, and exploits some of the IoT devices. Through this research we want to contribute to the existing literature on cyber-security that could potentially guide in both policy-making and intelligence research., Comment: 8 pages, 6 figures, 2 tables. This manuscript has been submitted to Springer for review (Manuscript ID: PDSE-D-24-00163) and is under consideration. It has not yet been peer-reviewed or published. Researchers are welcome to read and build upon this work; please cite it appropriately. For questions or clarifications, feel free to contact me

To develop trustworthy distributed systems, verification techniques and formal methods, including lightweight and practical approaches, have been employed to certify the design or implementation of security protocols. Lightweight formal methods offer a more accessible alternative to traditional fully formalised techniques by focusing on simplified models and tool support, making them more applicable in practical settings. The technical advantages of formal verification over manual testing are increasingly recognised in the cybersecurity community. However, for practitioners, formal modelling and verification are often too complex and unfamiliar to be used routinely. In this paper, we present an Eclipse IDE for the design, verification, and implementation of security protocols and evaluate its effectiveness, including feedback from users in educational settings. It offers user-friendly assistance in the formalisation process as part of a Model-Driven Development approach. This IDE centres around the Alice & Bob (AnB) notation, the AnBx Compiler and Code Generator, the OFMC model checker, and the ProVerif cryptographic protocol verifier. For the evaluation, we identify the six most prominent limiting factors for formal method adoption, based on relevant literature in this field, and we consider the IDE's effectiveness against those criteria. Additionally, we conducted a structured survey to collect feedback from university students who have used the toolkit for their projects. The findings demonstrate that this contribution is valuable as a workflow aid and helps users grasp essential cybersecurity concepts, even for those with limited knowledge of formal methods or cryptography. Crucially, users reported that the IDE has been an important component to complete their projects and that they would use again in the future, given the opportunity., Comment: 51 pages, 19 figures

In recent years, deepfakes (DFs) have been utilized for malicious purposes, such as individual impersonation, misinformation spreading, and artists' style imitation, raising questions about ethical and security concerns. However, existing surveys have focused on accuracy performance of passive DF detection approaches for single modalities, such as image, video or audio. This comprehensive survey explores passive approaches across multiple modalities, including image, video, audio, and multi-modal domains, and extend our discussion beyond detection accuracy, including generalization, robustness, attribution, and interpretability. Additionally, we discuss threat models for passive approaches, including potential adversarial strategies and different levels of adversary knowledge and capabilities. We also highlights current challenges in DF detection, including the lack of generalization across different generative models, the need for comprehensive trustworthiness evaluation, and the limitations of existing multi-modal approaches. Finally, we propose future research directions that address these unexplored and emerging issues in the field of passive DF detection, such as adaptive learning, dynamic benchmark, holistic trustworthiness evaluation, and multi-modal detectors for talking-face video generation., Comment: 26 pages

Graph Neural Networks (GNNs) have garnered significant attention from researchers due to their outstanding performance in handling graph-related tasks, such as social network analysis, protein design, and so on. Despite their widespread application, recent research has demonstrated that GNNs are vulnerable to backdoor attacks, implemented by injecting triggers into the training datasets. Trained on the poisoned data, GNNs will predict target labels when attaching trigger patterns to inputs. This vulnerability poses significant security risks for applications of GNNs in sensitive domains, such as drug discovery. While there has been extensive research into backdoor defenses for images, strategies to safeguard GNNs against such attacks remain underdeveloped. Furthermore, we point out that conventional backdoor defense methods designed for images cannot work well when directly implemented on graph data. In this paper, we first analyze the key difference between image backdoor and graph backdoor attacks. Then we tackle the graph defense problem by presenting a novel approach called MADE, which devises an adversarial mask generation mechanism that selectively preserves clean sub-graphs and further leverages masks on edge weights to eliminate the influence of triggers effectively. Extensive experiments across various graph classification tasks demonstrate the effectiveness of MADE in significantly reducing the attack success rate (ASR) while maintaining a high classification accuracy., Comment: 15 pages, 10 figures

The growing threat of deepfakes and manipulated media necessitates a radical rethinking of media authentication. Existing methods for watermarking synthetic data fall short, as they can be easily removed or altered, and current deepfake detection algorithms do not achieve perfect accuracy. Provenance techniques, which rely on metadata to verify content origin, fail to address the fundamental problem of staged or fake media. This paper introduces a groundbreaking paradigm shift in media authentication by advocating for the watermarking of real content at its source, as opposed to watermarking synthetic data. Our innovative approach employs multisensory inputs and machine learning to assess the realism of content in real-time and across different contexts. We propose embedding a robust realism score within the image metadata, fundamentally transforming how images are trusted and circulated. By combining established principles of human reasoning about reality, rooted in firmware and hardware security, with the sophisticated reasoning capabilities of contemporary machine learning systems, we develop a holistic approach that analyzes information from multiple perspectives. This ambitious, blue sky approach represents a significant leap forward in the field, pushing the boundaries of media authenticity and trust. By embracing cutting-edge advancements in technology and interdisciplinary research, we aim to establish a new standard for verifying the authenticity of digital media., Comment: Best Paper Award, Blue Sky Track at 26th ACM International Conference on Multimodal Interaction, Nov 2024, San Jose, Costa Rica