OutSafe-Bench: A Benchmark for Multimodal Offensive Content Detection in Large Language Models

Since Multimodal Large Language Models (MLLMs) are increasingly being integrated into everyday tools and intelligent agents, growing concerns have arisen regarding their possible output of unsafe contents, ranging from toxic language and biased imagery to privacy violations and harmful misinformation. Current safety benchmarks remain highly limited in both modality coverage and performance evaluations, often neglecting the extensive landscape of content safety. In this work, we introduce OutSafe-Bench, the first most comprehensive content safety evaluation test suite designed for the multimodal era. OutSafe-Bench includes a large-scale dataset that spans four modalities, featuring over 18,000 bilingual (Chinese and English) text prompts, 4,500 images, 450 audio clips and 450 videos, all systematically annotated across nine critical content risk categories. In addition to the dataset, we introduce a Multidimensional Cross Risk Score (MCRS), a novel metric designed to model and assess overlapping and correlated content risks across different categories. To ensure fair and robust evaluation, we propose FairScore, an explainable automated multi-reviewer weighted aggregation framework. FairScore selects top-performing models as adaptive juries, thereby mitigating biases from single-model judgments and enhancing overall evaluation reliability. Our evaluation of nine state-of-the-art MLLMs reveals persistent and substantial safety vulnerabilities, underscoring the pressing need for robust safeguards in MLLMs.

ULC: A Unified and Fine-Grained Controller for Humanoid Loco-Manipulation

Loco-Manipulation for humanoid robots aims to enable robots to integrate mobility with upper-body tracking capabilities. Most existing approaches adopt hierarchical architectures that decompose control into isolated upper-body (manipulation) and lower-body (locomotion) policies. While this decomposition reduces training complexity, it inherently limits coordination between subsystems and contradicts the unified whole-body control exhibited by humans. We demonstrate that a single unified policy can achieve a combination of tracking accuracy, large workspace, and robustness for humanoid loco-manipulation. We propose the Unified Loco-Manipulation Controller (ULC), a single-policy framework that simultaneously tracks root velocity, root height, torso rotation, and dual-arm joint positions in an end-to-end manner, proving the feasibility of unified control without sacrificing performance. We achieve this unified control through key technologies: sequence skill acquisition for progressive learning complexity, residual action modeling for fine-grained control adjustments, command polynomial interpolation for smooth motion transitions, random delay release for robustness to deploy variations, load randomization for generalization to external disturbances, and center-of-gravity tracking for providing explicit policy gradients to maintain stability. We validate our method on the Unitree G1 humanoid robot with 3-DOF (degrees-of-freedom) waist. Compared with strong baselines, ULC shows better tracking performance to disentangled methods and demonstrating larger workspace coverage. The unified dual-arm tracking enables precise manipulation under external loads while maintaining coordinated whole-body control for complex loco-manipulation tasks.

Voronoi-grid-based Pareto Front Learning and Its Application to Collaborative Federated Learning

Multi-objective optimization (MOO) exists extensively in machine learning, and aims to find a set of Pareto-optimal solutions, called the Pareto front, e.g., it is fundamental for multiple avenues of research in federated learning (FL). Pareto-Front Learning (PFL) is a powerful method implemented using Hypernetworks (PHNs) to approximate the Pareto front. This method enables the acquisition of a mapping function from a given preference vector to the solutions on the Pareto front. However, most existing PFL approaches still face two challenges: (a) sampling rays in high-dimensional spaces; (b) failing to cover the entire Pareto Front which has a convex shape. Here, we introduce a novel PFL framework, called as PHN-HVVS, which decomposes the design space into Voronoi grids and deploys a genetic algorithm (GA) for Voronoi grid partitioning within high-dimensional space. We put forward a new loss function, which effectively contributes to more extensive coverage of the resultant Pareto front and maximizes the HV Indicator. Experimental results on multiple MOO machine learning tasks demonstrate that PHN-HVVS outperforms the baselines significantly in generating Pareto front. Also, we illustrate that PHN-HVVS advances the methodologies of several recent problems in the FL field. The code is available at https://github.com/buptcmm/phnhvvs}{https://github.com/buptcmm/phnhvvs.

Implicit Jailbreak Attacks via Cross-Modal Information Concealment on Vision-Language Models

Multimodal large language models (MLLMs) enable powerful cross-modal reasoning capabilities. However, the expanded input space introduces new attack surfaces. Previous jailbreak attacks often inject malicious instructions from text into less aligned modalities, such as vision. As MLLMs increasingly incorporate cross-modal consistency and alignment mechanisms, such explicit attacks become easier to detect and block. In this work, we propose a novel implicit jailbreak framework termed IJA that stealthily embeds malicious instructions into images via least significant bit steganography and couples them with seemingly benign, image-related textual prompts. To further enhance attack effectiveness across diverse MLLMs, we incorporate adversarial suffixes generated by a surrogate model and introduce a template optimization module that iteratively refines both the prompt and embedding based on model feedback. On commercial models like GPT-4o and Gemini-1.5 Pro, our method achieves attack success rates of over 90% using an average of only 3 queries.

DD-Ranking: Rethinking the Evaluation of Dataset Distillation

In recent years, dataset distillation has provided a reliable solution for data compression, where models trained on the resulting smaller synthetic datasets achieve performance comparable to those trained on the original datasets. To further improve the performance of synthetic datasets, various training pipelines and optimization objectives have been proposed, greatly advancing the field of dataset distillation. Recent decoupled dataset distillation methods introduce soft labels and stronger data augmentation during the post-evaluation phase and scale dataset distillation up to larger datasets (e.g., ImageNet-1K). However, this raises a question: Is accuracy still a reliable metric to fairly evaluate dataset distillation methods? Our empirical findings suggest that the performance improvements of these methods often stem from additional techniques rather than the inherent quality of the images themselves, with even randomly sampled images achieving superior results. Such misaligned evaluation settings severely hinder the development of DD. Therefore, we propose DD-Ranking, a unified evaluation framework, along with new general evaluation metrics to uncover the true performance improvements achieved by different methods. By refocusing on the actual information enhancement of distilled datasets, DD-Ranking provides a more comprehensive and fair evaluation standard for future research advancements.

One Trigger Token Is Enough: A Defense Strategy for Balancing Safety and Usability in Large Language Models

Large Language Models (LLMs) have been extensively used across diverse domains, including virtual assistants, automated code generation, and scientific research. However, they remain vulnerable to jailbreak attacks, which manipulate the models into generating harmful responses despite safety alignment. Recent studies have shown that current safety-aligned LLMs often undergo the shallow safety alignment, where the first few tokens largely determine whether the response will be harmful. Through comprehensive observations, we find that safety-aligned LLMs and various defense strategies generate highly similar initial tokens in their refusal responses, which we define as safety trigger tokens. Building on this insight, we propose \texttt{D-STT}, a simple yet effective defense algorithm that identifies and explicitly decodes safety trigger tokens of the given safety-aligned LLM to trigger the model's learned safety patterns. In this process, the safety trigger is constrained to a single token, which effectively preserves model usability by introducing minimum intervention in the decoding process. Extensive experiments across diverse jailbreak attacks and benign prompts demonstrate that \ours significantly reduces output harmfulness while preserving model usability and incurring negligible response time overhead, outperforming ten baseline methods.

Interpretable Zero-shot Learning with Infinite Class Concepts

Zero-shot learning (ZSL) aims to recognize unseen classes by aligning images with intermediate class semantics, like human-annotated concepts or class definitions. An emerging alternative leverages Large-scale Language Models (LLMs) to automatically generate class documents. However, these methods often face challenges with transparency in the classification process and may suffer from the notorious hallucination problem in LLMs, resulting in non-visual class semantics. This paper redefines class semantics in ZSL with a focus on transferability and discriminability, introducing a novel framework called Zero-shot Learning with Infinite Class Concepts (InfZSL). Our approach leverages the powerful capabilities of LLMs to dynamically generate an unlimited array of phrase-level class concepts. To address the hallucination challenge, we introduce an entropy-based scoring process that incorporates a ``goodness" concept selection mechanism, ensuring that only the most transferable and discriminative concepts are selected. Our InfZSL framework not only demonstrates significant improvements on three popular benchmark datasets but also generates highly interpretable, image-grounded concepts. Code will be released upon acceptance.

ParetoHqD: Fast Offline Multiobjective Alignment of Large Language Models using Pareto High-quality Data

Aligning large language models with multiple human expectations and values is crucial for ensuring that they adequately serve a variety of user needs. To this end, offline multiobjective alignment algorithms such as the Rewards-in-Context algorithm have shown strong performance and efficiency. However, inappropriate preference representations and training with imbalanced reward scores limit the performance of such algorithms. In this work, we introduce ParetoHqD that addresses the above issues by representing human preferences as preference directions in the objective space and regarding data near the Pareto front as ''high-quality'' data. For each preference, ParetoHqD follows a two-stage supervised fine-tuning process, where each stage uses an individual Pareto high-quality training set that best matches its preference direction. The experimental results have demonstrated the superiority of ParetoHqD over five baselines on two multiobjective alignment tasks.

Local Data Quantity-Aware Weighted Averaging for Federated Learning with Dishonest Clients

Federated learning (FL) enables collaborative training of deep learning models without requiring data to leave local clients, thereby preserving client privacy. The aggregation process on the server plays a critical role in the performance of the resulting FL model. The most commonly used aggregation method is weighted averaging based on the amount of data from each client, which is thought to reflect each client's contribution. However, this method is prone to model bias, as dishonest clients might report inaccurate training data volumes to the server, which is hard to verify. To address this issue, we propose a novel secure \underline{Fed}erated \underline{D}ata q\underline{u}antity-\underline{a}ware weighted averaging method (FedDua). It enables FL servers to accurately predict the amount of training data from each client based on their local model gradients uploaded. Furthermore, it can be seamlessly integrated into any FL algorithms that involve server-side model aggregation. Extensive experiments on three benchmarking datasets demonstrate that FedDua improves the global model performance by an average of 3.17% compared to four popular FL aggregation methods in the presence of inaccurate client data volume declarations.

Token-Level Constraint Boundary Search for Jailbreaking Text-to-Image Models

Recent advancements in Text-to-Image (T2I) generation have significantly enhanced the realism and creativity of generated images. However, such powerful generative capabilities pose risks related to the production of inappropriate or harmful content. Existing defense mechanisms, including prompt checkers and post-hoc image checkers, are vulnerable to sophisticated adversarial attacks. In this work, we propose TCBS-Attack, a novel query-based black-box jailbreak attack that searches for tokens located near the decision boundaries defined by text and image checkers. By iteratively optimizing tokens near these boundaries, TCBS-Attack generates semantically coherent adversarial prompts capable of bypassing multiple defensive layers in T2I models. Extensive experiments demonstrate that our method consistently outperforms state-of-the-art jailbreak attacks across various T2I models, including securely trained open-source models and commercial online services like DALL-E 3. TCBS-Attack achieves an ASR-4 of 45\% and an ASR-1 of 21\% on jailbreaking full-chain T2I models, significantly surpassing baseline methods.