Large Vision-Language Models (LVLMs) have achieved substantial progress in cross-modal tasks. However, due to language bias, LVLMs are susceptible to object hallucination, which can be primarily divided into category, attribute, and relation hallucination, significantly impeding the trustworthy AI applications. Editing the internal activations of LVLMs has shown promising effectiveness in mitigating hallucinations with minimal cost. However, previous editing approaches neglect the effective guidance offered by factual textual semantics, thereby struggling to explicitly mitigate language bias. To address these issues, we propose Adaptive Factual-guided Visual-Textual Editing for hallucination mitigation (AFTER), which comprises Factual-Augmented Activation Steering (FAS) and Query-Adaptive Offset Optimization (QAO), to adaptively guides the original biased activations towards factual semantics. Specifically, FAS is proposed to provide factual and general guidance for activation editing, thereby explicitly modeling the precise visual-textual associations. Subsequently, QAO introduces a query-aware offset estimator to establish query-specific editing from the general steering vector, enhancing the diversity and granularity of editing. Extensive experiments on standard hallucination benchmarks across three widely adopted LVLMs validate the efficacy of the proposed AFTER, notably achieving up to a 16.3% reduction of hallucination over baseline on the AMBER benchmark. Our code and data will be released for reproducibility.

Applying Vision-Language Models (VLMs) to pervasive personal videos introduces profound privacy risks. This paper addresses the critical yet unexplored inferential privacy threat, specifically the risk of inferring sensitive personal attributes from seemingly benign data. To address this gap, we crowdsourced a dataset of 508 everyday personal videos from 58 individuals, and benchmarked VLM inference capabilities against human performance. Our findings reveal three key insights: (1) VLMs surpass recruited human evaluators in inferential accuracy, analyzing temporal behavioral patterns rather than relying solely on object recognition. (2) Inferential risk is strongly correlated with specific video characteristics and prompting strategies. (3) VLM-driven explanation towards the inference is unreliable, as we observe a disconnect between the model's reasoning and evidential impact, where ubiquitous objects often serve as misleading confounders.

Medical Large Vision-Language Models (Med-LVLMs) suffer from severe hallucinations, posing critical safety risks in clinical deployment. Editing LVLM activations has shown promise for mitigating hallucination with minimal cost. However, due to the requirements of medical domain expertise, existing methods struggle to capture imaging manifestations and diagnostic principles that are critical for clinical interpretation, thereby limiting their effectiveness. To address these limitations, we propose the first MEDical-oriented Activation Editing (MEDA) method by integrating Query-decisive Manifestation Steering (QMS) and Principle-driven Diagnosis Induction (PDI) to promote Med-LVLM's expertise elicitation. Specifically, QMS retrieves positive query-decisive imaging manifestations as trusted guidance for activation steering, while PDI constructs positive principle-embedded diagnostic prompts to induce expert-like clinical reasoning. Extensive experiments across six benchmarks and six LVLMs demonstrate that MEDA efficiently improves the response factuality with up to a 10.2\% gain on IU-Xray, while exhibiting strong generalization and few-shot robustness.

Algorithmic trading relies on machine learning models to make trading decisions. Despite strong in-sample performance, these models often degrade when confronted with evolving real-world market regimes, which can shift dramatically due to macroeconomic changes-e.g., monetary policy updates or unanticipated fluctuations in participant behavior. We identify two core challenges that perpetuate this mismatch: (1) insufficient robustness in existing policy against uncertainties in high-level market fluctuations, and (2) the absence of a realistic and diverse simulation environment for training, leading to policy overfitting. To address these issues, we propose a Bayesian Robust Framework that systematically integrates a macro-conditioned generative model with robust policy learning. On the data side, to generate realistic and diverse data, we propose a macro-conditioned GAN-based generator that leverages macroeconomic indicators as primary control variables, synthesizing data with faithful temporal, cross-instrument, and macro correlations. On the policy side, to learn robust policy against market fluctuations, we cast the trading process as a two-player zero-sum Bayesian Markov game, wherein an adversarial agent simulates shifting regimes by perturbing macroeconomic indicators in the macro-conditioned generator, while the trading agent-guided by a quantile belief network-maintains and updates its belief over hidden market states. The trading agent seeks a Robust Perfect Bayesian Equilibrium via Bayesian neural fictitious self-play, stabilizing learning under adversarial market perturbations. Extensive experiments on 9 financial instruments demonstrate that our framework outperforms 9 state-of-the-art baselines. In extreme events like the COVID pandemic, our method shows improved profitability and risk management, offering a reliable solution for trading under uncertain and rapidly shifting market dynamics.

Social media fact-checking has long been challenged by evidence-level and aggregation-level conflicts, where erroneous evidence mimics authoritative news sources. To capture this challenge and support conflict verification tasks, we present ContraNote, a large-scale real-world dataset curated from X's Community Notes system. It includes 33,686 posts for evaluating evidence-level conflict resolution, and 54,474 instances for evaluating aggregation-level prioritization. Additionally, we propose CoVer, a factual adjudication framework with three-stage pipelines: evidence schema normalization, factual consensus and support verification. This prioritizes evidence over noise to prevent it from compromising the final verdict. Technical evaluations show that CoVer achieves strong performance compared with state-of-the-art baselines across ContraNote (86.0

Flow matching has emerged as a standard paradigm for robotic manipulation owing to its strong expressive power for modelling complex, multimodal action distributions, alongside similar approaches like diffusion policy. However, existing methods rely on discretized action chunks, making them brittle to demonstrations collected at heterogeneous control frequencies and prone to temporally inconsistent actions that degrade control stability. In this paper, we propose Frequency-Aware Flow Matching (FAFM), which outputs continuous, temporally consistent actions. To handle heterogeneous frequency input, we transform discrete action sequences into the frequency domain with the discrete cosine transform (DCT), perform flow matching over the resulting coefficients, and reconstruct continuous actions via cosine basis expansion. To generate temporally consistent actions, we regularize the first-order temporal derivative to promote smooth actions. This corresponds to a Sobolev-type constraint that suppresses high-frequency errors and discourages abrupt action changes. Our FAFM is simple, introduces no additional network parameters and applies to standalone flow-matching policies and vision-language action models. Across synthetic toy benchmark, obstacle avoidance, LapGym, and LIBERO, FAFM improves success rates, multimodal expressivity, motion smoothness, convergence speed, robustness to mechanical bias and mixed-frequency input. These gains are consistent when deployed on a real-world Franka robot. Code available at https://anonymous.4open.science/r/FAFM.

In Vision–Language–Action (VLA) models, robustness to real-world perturbations is critical for deployment. Existing methods target simple visual disturbances, overlooking the broader multi-modal perturbations that arise in actions, instructions, environments, and observations. Here, we first evaluate the robustness of mainstream VLAs under 17 perturbations across four modalities. We find (1) actions as the most fragile modality, (2) Existing visual-robust VLA do not gain robustness in other modality, and (3) $\pi_0$ demonstrates superior robustness. To build multi-modal robust VLAs, we propose RobustVLA against perturbations in VLA inputs and outputs. For output robustness, we perform offline robust optimization against worst-case action noise that maximizes mismatch in flow matching objective. This can be seen as adversarial training, label smoothing, and outlier penalization. For input robustness, we enforce consistent actions across input variations that preserve task semantics. To account for multiple perturbations, we formulate robustness as a multi-armed bandit problem and apply an upper confidence bound algorithm to automatically identify the most harmful noise. Experiments on LIBERO demonstrate our RobustVLA delivers absolute gains over baselines of 12.6\% on the $\pi_0$ backbone and 10.4\% on the OpenVLA backbone across all 17 perturbations, achieving 50.6x faster inference than existing visual-robust BYOVLA that requires external LLMs, and a 10.4\% gain under mixed perturbations. On the real-world FR5 robot, under four types of multimodal perturbations, RobustVLA shows strong low-data performance, outperforming $\pi_0$ by $65.6\%$ success rate with 25 demonstrations. Even with abundant demos, our method still outperform $\pi_0$ by 30\% success rate. Code and demo videos available at \url{https://anonymous.4open.science/r/RobustVLA-283D}.

Partial agent failure becomes inevitable when systems scale up, making it crucial to identify the subset of agents whose failure causes worst-case system performance degradations. We study this Vulnerable Agent Identification (VAI) problem in large-scale multi-agent reinforcement learning (MARL). We frame VAI as a Hierarchical Adversarial Decentralized Mean Field Control (HAD-MFC), where where the upper level selects vulnerable agents as an NP-hard task and the lower level learns their worst-case adversarial policies via mean-field MARL. The two problems are coupled together, making HAD-MFC difficult to solve. To handle this, we first decouple the hierarchical process by Fenchel-Rockafellar transform, resulting a regularized mean-field Bellman operator for upper level that enables independent learning at each level, thus reducing computational complexity. We next reformulate the upper-level NP-hard problem as an MDP with dense rewards, allowing sequential identification of vulnerable agents via greedy and RL algorithms. This decomposition provably preserves the optimal solution. Experiments show our method effectively identifies more vulnerable agents in large-scale MARL and the rule-based system, fooling system into worse failures, and reveals the vulnerability of each agent in large systems. Code available at \url{https://anonymous.4open.science/r/VAI-5F61/}.

In cooperative multi-agent reinforcement learning (MARL), ensuring robustness against cooperative agents making unpredictable or worst-case adversarial actions is crucial for real-world deployment. In multi-agent settings, each agent may be perturbed or unperturbed, leading to an exponential increase in potential threat scenarios as the number of agents grows. Existing robust MARL methods either enumerate, or approximate all possible threat scenarios, leading to intense computation and insufficient robustness. In contrast, humans develop robust behaviors by maintaining a general level of caution rather than preparing for every possible threat. Inspired by human decision making, we frame robust MARL as a control-as-inference problem, and optimize worst-case robustness across all threat scenarios implicitly optimized through off-policy evaluation. Specifically, we introduce mutual information regularization as robust regularization (MIR3), which maximizes a lower bound on robustness during routine training, serving as a kind of caution for MARL without adversarial inputs. Further insights show that MIR3 acts as an information bottleneck, preventing agents from over-reacting to others and aligning policies with robust action priors. In the presence of worst-case adversaries, our MIR3 significantly surpasses baseline methods in robustness and training efficiency, and maintaining cooperative performance in StarCraft II, quadrotor swarm control, and robot swarm control. When deploying the robot swarm control algorithm in the real world, our method also outperforms the best baseline by 14.29% in reward. See code and demo videos at https://github.com/DIG-Beihang/MIR3

The proliferation of Vision-Language Models (VLMs) introduces profound privacy risks from personal videos. This paper addresses the critical yet unexplored inferential privacy threat, the risk of inferring sensitive personal attributes over the data. To address this gap, we crowdsourced a dataset of 508 everyday personal videos from 58 individuals. We then conducted a benchmark study evaluating VLM inference capabilities against human performance. Our findings reveal three critical insights: (1) VLMs possess superhuman inferential capabilities, significantly outperforming human evaluators, leveraging a shift from object recognition to behavioral inference from temporal streams. (2) Inferential risk is strongly correlated with factors such as video characteristics and prompting strategies. (3) VLM-driven explanation towards the inference is unreliable, as we revealed a disconnect between the model-generated explanations and evidential impact, identifying ubiquitous objects as misleading confounders.

The reasoning capabilities of embodied agents introduce a critical, under-explored inferential privacy challenge, where the risk of an agent generate sensitive conclusions from ambient data. This capability creates a fundamental tension between an agent's utility and user privacy, rendering traditional static controls ineffective. To address this, this position paper proposes a framework that reframes privacy as a dynamic learning problem grounded in theory of Contextual Integrity (CI). Our approach enables agents to proactively learn and adapt to individual privacy norms through interaction, outlining a research agenda to develop embodied agents that are both capable and function as trustworthy safeguards of user privacy.

The proliferation of AI agents, with their complex and context-dependent actions, renders conventional privacy paradigms obsolete. This position paper argues that the current model of privacy management, rooted in a user's unilateral control over a passive tool, is inherently mismatched with the dynamic and interactive nature of AI agents. We contend that ensuring effective privacy protection necessitates that the agents proactively align with users' privacy preferences instead of passively waiting for the user to control. To ground this shift, and using personalized conversational recommendation agents as a case, we propose a conceptual framework built on Contextual Integrity (CI) theory and Privacy Calculus theory. This synthesis first reframes automatically controlling users' privacy as an alignment problem, where AI agents initially did not know users' preferences, and would learn their privacy preferences through implicit or explicit feedback. Upon receiving the preference feedback, the agents used alignment and Pareto optimization for aligning preferences and balancing privacy and utility. We introduced formulations and instantiations, potential applications, as well as five challenges.

Goal assignment is a critical challenge in multi-robot systems. The emergence of large language models (LLMs) has enabled the use of natural language commands for tackling goal assignment problems. However, applying LLMs directly to these tasks presents two limitations: 1) limited accuracy and 2) excessive decision delays due to their autoregressive nature, hindering adaptability to unexpected changes. To address these issues, inspired by dual-process theory, we propose a framework called Collaborative LLMs for dynamic Goal Assignment (CLGA). Specifically, we leverage LLMs for pre-planning tasks and invoke an external solver to generate an initial goal assignment solution, ensuring solution accuracy. During execution, small-scale models enable real-time adjustments to respond to dynamic environmental changes. This approach integrates the strengths of slow, precise pre-planning and fast, adaptive online adjustments, allowing agents to efficiently handle real-world challenges. Additionally, we introduce a benchmark dataset for NLP-based goal assignment to advance research in this domain. Simulation and real-world experiments demonstrate that CLGA significantly enhances task execution efficiency and flexibility in multi-robot systems. The prompt, experimental videos, and datasets associated with this work are available at https://sites.google.com/view/project-clga/.

This study probes the vulnerabilities of cooperative multi-agent reinforcement learning (c-MARL) under adversarial attacks, a critical determinant of c-MARL's worst-case performance prior to real-world implementation. Current observation-based attacks, constrained by white-box assumptions, overlook c-MARL's complex multi-agent interactions and cooperative objectives, resulting in impractical and limited attack capabilities. To address these shortcomes, we propose Adversarial Minority Influence (AMI), a practical and strong for c-MARL. AMI is a practical black-box attack and can be launched without knowing victim parameters. AMI is also strong by considering the complex multi-agent interaction and the cooperative goal of agents, enabling a single adversarial agent to unilaterally misleads majority victims to form targeted worst-case cooperation. This mirrors minority influence phenomena in social psychology. To achieve maximum deviation in victim policies under complex agent-wise interactions, our unilateral attack aims to characterize and maximize the impact of the adversary on the victims. This is achieved by adapting a unilateral agent-wise relation metric derived from mutual information, thereby mitigating the adverse effects of victim influence on the adversary. To lead the victims into a jointly detrimental scenario, our targeted attack deceives victims into a long-term, cooperatively harmful situation by guiding each victim towards a specific target, determined through a trial-and-error process executed by a reinforcement learning agent. Through AMI, we achieve the first successful attack against real-world robot swarms and effectively fool agents in simulated environments into collectively worst-case scenarios, including Starcraft II and Multi-agent Mujoco. The source code and demonstrations can be found at: https://github.com/DIG-Beihang/AMI.

In cooperative Multi-Agent Reinforcement Learning (MARL), it is a common practice to tune hyperparameters in ideal simulated environments to maximize cooperative performance. However, policies tuned for cooperation often fail to maintain robustness and resilience under real-world uncertainties. Building trustworthy MARL systems requires a deep understanding of \emph{robustness}, which ensures stability under uncertainties, and \emph{resilience}, the ability to recover from disruptions—a concept extensively studied in control systems but largely overlooked in MARL. In this paper, we present a large-scale empirical study comprising over 82,620 experiments to evaluate cooperation, robustness, and resilience in MARL across 4 real-world environments, 13 uncertainty types, and 15 hyperparameters. Our key findings are: (1) Under mild uncertainty, optimizing cooperation improves robustness and resilience, but this link weakens as perturbations intensify. Robustness and resilience also varies by algorithm and uncertainty type. (2) Robustness and resilience do not generalize across uncertainty modalities or agent scopes: policies robust to action noise for all agents may fail under observation noise on a single agent. (3) Hyperparameter tuning is critical for trustworthy MARL: surprisingly, standard practices like parameter sharing, GAE, and PopArt can hurt robustness, while early stopping, high critic learning rates, and Leaky ReLU consistently help. By optimizing hyperparameters only, we observe substantial improvement in cooperation, robustness and resilience across all MARL backbones, with the phenomenon also generalizing to robust MARL methods across these backbones.

As intelligence increases, so does its shadow. AI deception, in which systems induce false beliefs to secure self-beneficial outcomes, has evolved from a speculative concern to an empirically demonstrated risk across language models, AI agents, and emerging frontier systems. This project provides a comprehensive and up-to-date overview of the AI deception field, covering its core concepts, methodologies, genesis, and potential mitigations. First, we identify a formal definition of AI deception, grounded in signaling theory from studies of animal deception. We then review existing empirical studies and associated risks, highlighting deception as a sociotechnical safety challenge. We organize the landscape of AI deception research as a deception cycle, consisting of two key components: deception emergence and deception treatment. Deception emergence reveals the mechanisms underlying AI deception: systems with sufficient capability and incentive potential inevitably engage in deceptive behaviors when triggered by external conditions. Deception treatment, in turn, focuses on detecting and addressing such behaviors. On deception emergence, we analyze incentive foundations across three hierarchical levels and identify three essential capability preconditions required for deception. We further examine contextual triggers, including supervision gaps, distributional shifts, and environmental pressures. On deception treatment, we conclude detection methods covering benchmarks and evaluation protocols in static and interactive settings. Building on the three core factors of deception emergence, we outline potential mitigation strategies and propose auditing approaches that integrate technical, community, and governance efforts to address sociotechnical challenges and future AI risks. To support ongoing work in this area, we release a living resource at www.deceptionsurvey.com.

In this study, we explore the robustness of cooperative multi-agent reinforcement learning (c-MARL) against Byzantine failures, where any agent can enact arbitrary, worst-case actions due to malfunction or adversarial attack. To address the uncertainty that any agent can be adversarial, we propose a Bayesian Adversarial Robust Dec-POMDP (BARDec-POMDP) framework, which views Byzantine adversaries as nature-dictated types, represented by a separate transition. This allows agents to learn policies grounded on their posterior beliefs about the type of other agents, fostering collaboration with identified allies and minimizing vulnerability to adversarial manipulation. We define the optimal solution to the BARDec-POMDP as an ex post robust Bayesian Markov perfect equilibrium, which we proof to exist and weakly dominates the equilibrium of previous robust MARL approaches. To realize this equilibrium, we put forward a two-timescale actor-critic algorithm with almost sure convergence under specific conditions. Experimentation on matrix games, level-based foraging and StarCraft II indicate that, even under worst-case perturbations, our method successfully acquires intricate micromanagement skills and adaptively aligns with allies, demonstrating resilience against non-oblivious adversaries, random allies, observation-based attacks, and transfer-based attacks.

Billions of people share images from their daily lives on social media every day. However, their biometric information ( e.g ., fingerprints) could be easily stolen from these images. The threat of fingerprint leakage from social media has created a strong desire to anonymize shared images while maintaining image quality, since fingerprints act as a lifelong individual biometric password. To guard the fingerprint leakage, adversarial attack that involves adding imperceptible perturbations to fingerprint images have emerged as a feasible solution. However, existing works of this kind are either weak in black-box transferability or cause the images to have an unnatural appearance. Motivated by the visual perception hierarchy ( i.e ., high-level perception exploits model-shared semantics that transfer well across models while low-level perception extracts primitive stimuli that result in high visual sensitivity when a suspicious stimulus is provided), we propose FingerSafe, a hierarchical perceptual protective noise injection framework to address the above mentioned problems. For black-box transferability, we inject protective noises into the fingerprint orientation field to perturb the model-shared high-level semantics ( i.e ., fingerprint ridges). Considering visual naturalness, we suppress the low-level local contrast stimulus by regularizing the response of the Lateral Geniculate Nucleus. Our proposed FingerSafe is the first to provide feasible fingerprint protection in both digital (up to 94.12%) and realistic scenarios (Twitter and Facebook, up to 68.75%). Our code can be found at https://github.com/nlsde-safety-team/FingerSafe.

Reinforcement learning (RL) has shown great promise in addressing multi-agent collision avoidance challenges. However, existing RL-based methods often suffer from low training efficiency and poor action safety. To tackle these issues, we introduce a physics-informed reinforcement learning framework equipped with two modules: a Potential Field (PF) module and a Multi-Agent Multi-Level Safety (MAMLS) module. The PF module uses the Artificial Potential Field method to compute a regularization loss, adaptively integrating it into the critic's loss to enhance training efficiency. The MAMLS module formulates action safety as a constrained optimization problem, deriving safe actions by solving this optimization. Furthermore, to better address the characteristics of multi-agent collision avoidance tasks, multi-agent multi-level constraints are introduced. The results of simulations and real-world experiments showed that our physics-informed framework offers a significant improvement in terms of both the efficiency of training and safety-related metrics over advanced baseline methods.