Federated Recommender Systems (FedRecs) leverage federated learning to protect user privacy by retaining data locally. However, user embeddings in FedRecs often encode sensitive attribute information, rendering them vulnerable to attribute inference attacks. Attribute unlearning has emerged as a promising approach to mitigate this issue. In this paper, we focus on user-level FedRecs, which is a more practical yet challenging setting compared to group-level FedRecs. Adversarial training emerges as the most feasible approach within this context. We identify two key challenges in implementing adversarial training-based attribute unlearning for user-level FedRecs: i) mitigating training instability caused by user data heterogeneity, and ii) preventing attribute information leakage through gradients. To address these challenges, we propose FedAU2, an attribute unlearning method for user-level FedRecs. For CH1, we propose an adaptive adversarial training strategy, where the training dynamics are adjusted in response to local optimization behavior. For CH2, we propose a dual-stochastic variational autoencoder to perturb the adversarial model, effectively preventing gradient-based information leakage. Extensive experiments on three real-world datasets demonstrate that our proposed FedAU2 achieves superior performance in unlearning effectiveness and recommendation performance compared to existing baselines.

Recent advancements in large language models (LLMs) have significantly enhanced the capabilities of collaborative multi-agent systems, enabling them to address complex challenges. However, within these multi-agent systems, the susceptibility of agents to collective cognitive biases remains an underexplored issue. A compelling example is the Mandela effect, a phenomenon where groups collectively misremember past events as a result of false details reinforced through social influence and internalized misinformation. This vulnerability limits our understanding of memory bias in multi-agent systems and raises ethical concerns about the potential spread of misinformation. In this paper, we conduct a comprehensive study on the Mandela effect in LLM-based multi-agent systems, focusing on its existence, causing factors, and mitigation strategies. We propose ManBench, a novel benchmark designed to evaluate agent behaviors across four common task types that are susceptible to the Mandela effect, using five interaction protocols that vary in agent roles and memory timescales. We evaluate agents powered by several LLMs on ManBench to quantify the Mandela effect, and analyze how different factors affect it. Moreover, we propose strategies to mitigate this effect, including prompt-level defenses (e.g., cognitive anchoring and source scrutiny) and model-level alignment-based defense, achieving an average 74.40% reduction in the Mandela effect compared to the baseline. Our findings provide valuable insights for developing more resilient and ethically aligned collaborative multi-agent systems.

Large Language Models (LLMs) have been integrated into complex ecosystems (e.g., Code Agents), while Indirect Prompt Injection (IPI) attacks have emerged as critical barriers to their safe deployment. Attackers exploit LLMs' indistinguishability between "instructions" and "data" to manipulate LLMs via maliciously injected instructions. Existing defenses, however, face an intractable safety-utility trade-off: most guardrails either incur high latency or suffer from severe over-refusal. In this paper, we first demonstrate that LLMs can separate instruction from data intrinsically with both theoretical and empirical evidence. Inspired by this insight, we propose AEGIS (Adaptive Ensemble Guard for Injection Shielding). AEGIS extracts instruction-sensitive projectors to identify malicious instructions and leverages a Unified Multi-Layer Consensus mechanism that aggregates topologically distinct signals across the network depth. Empirical evaluations show that AEGIS achieves remarkable detection performance against both heuristic and optimization-based attacks compared to baselines, highlighting its potential to mitigate IPI. Code is available at https://github.com/xaddwell/AEGIS

Large vision-language models (LVLMs) have achieved remarkable advancements in multimodal reasoning tasks. However, their widespread accessibility raises critical concerns about potential copyright infringement. Will LVLMs accurately recognize and comply with copyright regulations when encountering copyrighted content (i.e., user input, retrieved documents) in the context? Failure to comply with copyright regulations may lead to serious legal and ethical consequences, particularly when LVLMs generate responses based on copyrighted materials (e.g., retrieved book experts, news reports). In this paper, we present a comprehensive evaluation of various LVLMs, examining how they handle copyrighted content – such as book excerpts, news articles, music lyrics, and code documentation when they are presented as visual inputs. To systematically measure copyright compliance, we introduce a large-scale benchmark dataset comprising 50,000 multimodal query-content pairs designed to evaluate how effectively LVLMs handle queries that could lead to copyright infringement. Given that real-world copyrighted content may or may not include a copyright notice, the dataset includes query-content pairs in two distinct scenarios: with and without a copyright notice. For the former, we extensively cover four types of copyright notices to account for different cases. Our evaluation reveals that even state-of-the-art closed-source LVLMs exhibit significant deficiencies in recognizing and respecting the copyrighted content, even when presented with the copyright notice. To solve this limitation, we introduce a novel tool-augmented defense framework for copyright compliance, which reduces infringement risks in all scenarios. Our findings underscore the importance of developing copyright-aware LVLMs to ensure the responsible and lawful use of copyrighted content.

Safety-aligned large language models (LLMs) are increasingly deployed in real-world pipelines, yet this deployment also enlarges the supply-chain attack surface: adversaries can distribute backdoored checkpoints that behave normally under standard evaluation but jailbreak when a hidden trigger is present. Recent post-hoc weight-editing methods offer an efficient approach to injecting such backdoors by directly modifying model weights to map a trigger to an attacker-specified response. However, existing methods typically optimize a token-level mapping that forces an affirmative prefix (e.g., “Sure”), which does not guarantee sustained harmful output—the model may begin with apparent agreement yet revert to safety-aligned refusal within a few decoding steps. We address this reliability gap by shifting the backdoor objective from surface tokens to internal representations. We extract a steering vector that captures the difference between compliant and refusal behaviors, and compile it into a persistent weight modification that activates only when the trigger is present. To preserve stealthiness and benign utility, we impose a null-space constraint so that the injected edit remains dormant on clean inputs. The method is efficient, requiring only a small set of examples and admitting a closed-form solution. Across multiple safety-aligned LLMs and jailbreak benchmarks, our method achieves high triggered attack success while maintaining non-triggered safety and general utility.

The prosperity of text-to-image (T2I) models has fostered a vibrant share-and-play ecosystem centered on Low-Rank Adaptation (LoRA) plugins, which allow users to customize and share model capabilities with ease. This democratization, however, comes with a hidden but severe security risk. Malicious users could share and distribute seemingly benign LoRA plugins that contain hidden functionalities to poison the model-sharing market, like Civitai or Liblib, severely undermining the user trust that underpins this collaborative ecosystem and threatening the safety of countless downstream applications. Despite these risks, plugin poisoning in the real-world T2I ecosystem remains underexplored. This paper introduces PoisonLoRA, the first systematic study of LoRA plugin supply-chain risks that exploits the trust and characteristics within the T2I ecosystem. We identify two primary attack instances: (1) Concept Hijacking, where a hijacked LoRA could generate images to influence public opinion and spread propaganda, and (2) Task Injection, where a LoRA is injected to produce harmful content (e.g., NSFW images) only activated by a secret key. Critically, the malicious payload persists with virus-like propagation. Such propagations weaponize the very act of creative collaboration (e.g., LoRA merging) to spread its contagion, turning every remix into a new carrier. Extensive experiments validate that PoisonLoRA is both effective and stealthy. Specifically, we achieve approximately 100

Machine-learning services increasingly rely on public data, third-party providers, and outsourced training, creating opportunities for data-poisoning attacks that implant persistent malicious behavior while preserving benign utility. However, existing backdoor studies largely evaluate exact trigger reuse, training-exposed trigger diversity, or variations along predefined transformation axes. They therefore leave a critical blind spot: whether a backdoor learned from one training-time trigger can generalize to an inference-time trigger family absent from victim training. We formulate this problem as backdoor generalization under training–inference trigger shift and introduce Lilith, a black-box anchor-to-family framework. Using only disjoint surrogate resources, Lilith first induces a compact target-side vulnerability with a single training anchor, then constructs a bounded inference-only family that preserves the anchor-induced representation geometry. We characterize this mechanism through anchor clearance and family reach, deriving sufficient conditions for family-wise target preservation under local regularity and bounded surrogate–victim discrepancy. Experiments across datasets, architectures, poisoning rates, and defenses show that Lilith achieves high family-wise attack success with limited utility degradation and a small trigger generalization gap. Additional analyses show that family activation depends on representation alignment rather than the proposal mechanism, exposing a broader threat overlooked by exact-trigger evaluation.

Multimodal guard models have emerged as critical safety components for screening content in vision-language systems. While adversarial research has extensively studied jailbreaking attacks that produce false negatives, the inverse threat of inducing false positives on benign inputs remains unexplored. We introduce Unsafe Induction Attacks, where adversaries distribute imperceptibly perturbed safe images that trigger guard models to reject legitimate user requests, causing a "Boy Who Cried Wolf" effect that degrades service availability and erodes trust. This reveals an availability failure mode in deployed safety filters. To realize this threat under diverse user prompts, we propose Unsafe Semantic Distillation (USD), which aligns adversarial perturbations with distributional representations of unsafe content rather than prompt-specific instances. Evaluated on four state-of-the-art guard models across realistic user simulation scenarios, USD achieves 84

Coding agents are increasingly used for software engineering tasks, including bootstrapping projects from third-party repositories whose integrity cannot be assumed. Prior work on repository poisoning largely focuses on attacker-controlled injection and disguise, but developers also shape risk through everyday invocation choices: what task to delegate, how to phrase the request, and which skills or rules to supply. We term these user-side choices Prompt-Level Configurations (PLCs) and introduce CIPR (Coding In Poisoned Repos), the first benchmark that systematically varies PLCs in poisoned real-world repositories. CIPR comprises 1,920 instances across 20 repositories, four task types, three social-media-grounded prompt styles, and three skill/rule conditions, and measures attack success rate (ASR) and agent alert rate (AR) using automated runtime and trace-based oracles. Our evaluation reveals two key insights: (1) Vulnerability is highly context-dependent, with task type creating up to a 4.5-fold difference in ASR, with test-execution task forming a silent attack surface (high ASR, low AR). (2) Prompt expression shifts risk indirectly: underspecified prompts reduce ASR by truncating execution depth; noisy prompts exhibit a directional trend toward suppressing alerts by making malicious content less conspicuous. These findings highlight that coding agent vulnerability is not a static property, but a dynamic outcome shaped by everyday user configurations.

Federated learning (FL) enables collaborative model training across distributed nodes without exposing raw data, but its decentralized nature makes it vulnerable in trust-deficient environments. Inference attacks may recover sensitive information from gradient updates, while poisoning attacks can degrade model performance or induce malicious behaviors. Existing defenses often suffer from high communication and computation costs, or limited detection precision. To address these issues, we propose LSHFed, a robust and communication-efficient FL framework that simultaneously enhances aggregation robustness and privacy preservation. At its core, LSHFed incorporates LSHGM, a novel gradient verification mechanism that projects high-dimensional gradients into compact binary representations via multi-hyperplane locality-sensitive hashing. This enables accurate detection and filtering of malicious gradients using only their irreversible hash forms, thus mitigating privacy leakage risks and substantially reducing transmission overhead. Extensive experiments demonstrate that LSHFed maintains high model performance even when up to 50% of participants are collusive adversaries, while achieving up to a 1000× reduction in gradient verification communication compared to full-gradient methods.

Large language models (LLMs) have been widely integrated into critical automated workflows, including contract review and job application processes. However, LLMs are susceptible to manipulation by fraudulent information, which can lead to harmful outcomes. Although advanced defense methods have been developed to address this issue, they often exhibit limitations in effectiveness, interpretability, and generalizability, particularly when applied to LLM-based applications. To address these challenges, we introduce FraudShield, a novel framework designed to protect LLMs from fraudulent content by leveraging a comprehensive analysis of fraud tactics. Specifically, FraudShield constructs and refines a fraud tactic-keyword knowledge graph to capture high-confidence associations between suspicious text and fraud techniques. The structured knowledge graph augments the original input by highlighting keywords and providing supporting evidence, guiding the LLM toward more secure responses. Extensive experiments show that FraudShield consistently outperforms state-of-the-art defenses across four mainstream LLMs and five representative fraud types, while also offering interpretable clues for the model's generations.

Large language model (LLM)-based applications rely on system prompts to encode core logic and developer-defined constraints, making these prompts important intellectual property. However, system prompts are vulnerable to prompt leaking attacks. Although prior work has shown such attacks in controlled settings, their prevalence, causes, and defenses in real-world deployments remain unclear. This paper presents a systematic study of prompt leaking in real-world LLM-based applications. We measure 1,200 applications across six major commercial platforms and find that over 80

Puns are a common form of rhetorical wordplay that exploits polysemy and phonetic similarity to create humor. In multimodal puns, visual and textual elements synergize to ground the literal sense and evoke the figurative meaning simultaneously. Although Vision-Language Models (VLMs) are widely used in multimodal understanding and generation, their ability to understand puns has not been systematically studied due to a scarcity of rigorous benchmarks. To address this, we first propose a multimodal pun generation pipeline. We then introduce MultiPun, a dataset comprising diverse types of puns alongside adversarial non-pun distractors. Our evaluation reveals that most models struggle to distinguish genuine puns from these distractors. Moreover, we propose both prompt-level and model-level strategies to enhance pun comprehension, with an average improvement of 16.5% in F1 scores. Our findings provide valuable insights for developing future VLMs that master the subtleties of human-like humor via cross-modal reasoning.

Large Language Models (LLMs) have been widely applied in high-stakes decision-making scenarios such as corporate strategy, and users are increasingly relying on their outputs. However, the deep integration of open-source model sharing ecosystems with LLM-powered critical decision-making applications also introduces critical risks: if an attacker can manipulate the model's cognitive stance, they can indirectly influence the judgments and actions of downstream decision-makers. This paper defines such threats as decision-level hijacking. Existing attacks fail to achieve targeted cognitive manipulation without triggering prohibited content or degrading model functionality. To fill this gap, this paper reveals that Bit-Flip Attacks (BFAs) can serve as an attack vector for inducing decision-level hijacking, requiring no real-time interaction or control over the training process, and only a minimal number of weight bits need to be flipped after deployment to achieve stealthy, low-cost, and persistent cognitive manipulation. Therefore, we propose CogBias, a cognitive bias injection framework for LLMs. CogBias converts subjective preferences into optimization signals via a differentiable sentiment evaluator, uses a multi-objective loss to jointly constrain multiple dimensions, and constructs BitScout to locate critical bits, achieving targeted cognitive intervention under an ultra-sparse flip budget. Experiments on Llama-3.2-3B, Mistral-7B, and Qwen2.5-14B, as well as on the commercial recommendation and controversial factual topic scenarios, demonstrate that flipping only a small number of bits stably induces significant stance shifts on target topics, while the impact on non-target tasks and overall output distribution is limited. This work demonstrates that minute perturbations to low-level weight data suffice to undermine the high-level value alignment of LLMs.

Federated Learning (FL) offers collaborative model training across multiple decentralized devices without the need to share data directly, enhancing privacy and data security. However, FL systems are susceptible to backdoor attacks, where malicious clients inject poisoned weights during training. Existing defenses, primarily based on anomaly detection, are prone to erroneous rejections of normal weights while accepting poisoned ones, largely due to shortcomings in quantifying similarities among client models. Furthermore, other defenses demonstrate effectiveness only when dealing with a limited number of malicious clients, typically fewer than 10%. To alleviate these vulnerabilities, we present G2uardFL, a protective framework that translates the detection of malicious clients into an attributed graph clustering problem, thus safeguarding FL systems. Specifically, this framework employs a client graph clustering approach to identify malicious clients and integrates an adaptive mechanism to amplify the discrepancy between the aggregated model and the poisoned ones, effectively eliminating embedded backdoors. Through empirical evaluation, comparing G2uardFL with cutting-edge defenses, such as FLAME (USENIX Security 2022) and DeepSight (NDSS 2022), against various backdoor attacks, including 3DFed (SP 2023), our results demonstrate its significant effectiveness in mitigating backdoor attacks while having a negligible impact on the aggregated model's performance on benign samples (i.e., the primary task performance). For instance, in an FL system with 25% malicious clients, G2uardFL reduces the attack success rate to 10.61%, while maintaining a primary task performance of 80.98% on the CIFAR-10 dataset. This surpasses the performance of the best-performing baseline, which merely achieves the attack success rate of 19.54%.

Dataset distillation (DD) compresses large datasets into smaller ones while preserving the performance of models trained on them. Although DD is often assumed to enhance data privacy by aggregating over individual examples, recent studies reveal that standard DD can still leak sensitive information from the original dataset due to the lack of formal privacy guarantees. Existing differentially private (DP)-DD methods attempt to mitigate this risk by injecting noise into the distillation process. However, they often fail to fully leverage the original dataset, resulting in degraded realism and utility. This paper introduces DP-GENG, a novel framework that addresses the key limitations of current DP-DD by leveraging DP-generated data. Specifically, DP-GENG initializes the distilled dataset with DP-generated data to enhance realism. Then, generated data refines the DP-feature matching technique to distill the original dataset under a small privacy budget, and trains an expert model to align the distilled examples with their class distribution. Furthermore, we design a privacy budget allocation strategy to determine budget consumption across DP components and provide a theoretical analysis of the overall privacy guarantees. Extensive experiments show that DP-GENG significantly outperforms state-of-the-art DP-DD methods in terms of both dataset utility and robustness against membership inference attacks, establishing a new paradigm for privacy-preserving dataset distillation.

The proliferation of Low-Rank Adaptation (LoRA) models has democratized personalized text-to-image generation, enabling users to share lightweight models (e.g., personal portraits) on platforms like Civitai and Liblib. However, this "share-and-play" ecosystem introduces critical risks: benign LoRAs can be weaponized by adversaries to generate harmful content (e.g., political, defamatory imagery), undermining creator rights and platform safety. Existing defenses like concept-erasure methods focus on full diffusion models (DMs), neglecting LoRA's unique role as a modular adapter and its vulnerability to adversarial prompt engineering. To bridge this gap, we propose LoRAShield, the first data-free editing framework for securing LoRA models against misuse. Our platform-driven approach dynamically edits and realigns LoRA's weight subspace via adversarial optimization and semantic augmentation. Experimental results demonstrate that LoRAShield achieves remarkable effectiveness, efficiency, and robustness in blocking malicious generations without sacrificing the functionality of the benign task. By shifting the defense to platforms, LoRAShield enables secure, scalable sharing of personalized models, a critical step toward trustworthy generative ecosystems.

Large Language Models (LLMs) have revolutionized how information are collected, aggregated, and reasoned. However, this enables a novel and accessible vector of privacy intrusion: the automated and in-depth personal profiling; this engenders a chilling effect of "peepers everywhere". Existing research primarily unfolds from the training pipeline of LLM, emphasizing the exposure of Personally Identifiable Information (PII) through memorization, while privacy studies from a human-centric perspective remain underexplored. To fill this void, we empirically investigate privacy perception in the real world through the lens of human awareness and the practices of LLM-integrated platforms, revealing a significant dissonance: platforms fail to technically or policy-wise address public privacy concerns. To facilitate a systematic and quantifiable study of privacy risk, we propose the PrivacyIceberg, which categorizes real-world human privacy risks into three tiers: explicitly searched, contextually inferred, and deeply aggregated, based on the sophistication of LLM exploitation. We developed IcebergExplorer to audit privacy exposure, utilizing minimal PII as a search seed to reconstruct high-fidelity profiles, achieving over 90% factual accuracy within 10 minutes at a cost under $3, for real-world scenarios. Additionally, we identify six root causes contributing to such privacy disclosures and propose multi-stakeholder countermeasures for LLM vendors, individuals, and data publishers.

Deep neural networks (DNNs) underpin critical applications yet remain vulnerable to backdoor attacks, typically reliant on heuristic brute-force methods. Despite significant empirical advancements in backdoor research, the lack of rigorous theoretical analysis limits understanding of underlying mechanisms, constraining attack predictability and adaptability. Therefore, we provide a theoretical analysis targeting backdoor attacks, focusing on how sparse decision boundaries enable disproportionate model manipulation. Based on this finding, we derive a closed-form "ambiguous boundary region" wherein negligible relabeled samples induce substantial misclassification. Influence function analysis further quantifies significant parameter shifts caused by these margin samples, with minimal impact on clean accuracy, formally grounding why such low poison rates suffice for efficacious attacks. Leveraging these insights, we propose Eminence, an explainable and robust black-box backdoor framework with provable theoretical guarantees and inherent stealth properties. Eminence optimizes a universal, visually subtle trigger that strategically exploits vulnerable decision boundaries and effectively achieves robust misclassification with exceptionally low poison rates (<= 0.01%, compared to SOTA methods typically requiring >= 1%). Comprehensive experiments validate our theoretical discussions and demonstrate the effectiveness of Eminence, confirming an exponential relationship between margin poisoning and adversarial boundary manipulation. Eminence maintains >= 90% attack success rate, exhibits negligible clean-accuracy loss, and demonstrates high transferability across diverse models, datasets and scenarios. Our code is available at https://github.com/NESA-Lab/Eminence

Artificial Intelligence (AI) agents have evolved from passive predictive tools into active entities capable of autonomous decision-making and environmental interaction, driven by the reasoning capabilities of Large Language Models (LLMs). However, this evolution has introduced critical security vulnerabilities that existing frameworks fail to address. The Hierarchical Autonomy Evolution (HAE) framework organizes agent security into three tiers: Cognitive Autonomy (L1) targets internal reasoning integrity; Execution Autonomy (L2) covers tool-mediated environmental interaction; Collective Autonomy (L3) addresses systemic risks in multi-agent ecosystems. We present a taxonomy of threats spanning cognitive manipulation, physical environment disruption, and multi-agent systemic failures, and evaluate existing defenses while identifying key research gaps. The findings aim to guide the development of multilayered, autonomy-aware defense architectures for trustworthy AI agent systems.