arXiv subjects

Bohan Jiang

Publications and source records attributed to Bohan Jiang.

At least 19 recordsLinked to original sources

arXiv2026

The Customer Never Sees the Goal: Measuring and Detecting Vulnerability-Targeted Manipulation in LLM Assistants

Companies deploy large language model (LLM) assistants on their websites to advise customers. For commercial reasons, the operator can give an assistant a goal that the customer never sees. This paper focuses on operator-induced vulnerability-targeted manipulation (VTM), in which an assistant pursues such a goal by exploiting a vulnerability the customer discloses. While existing research mainly measures how manipulative a model is, we go further and ask whether VTM can be detected from the customer-AI conversation. To investigate this at scale, we introduce GAP (Goal-Anchored Probing). GAP first records the option a model recommends without any goal and then sets the other option as the operator's goal. We apply GAP to 15 open- and closed-source LLMs and collect 619,600 labeled replies across 540 everyday scenarios in 12 domains. We organize our study around three research questions. (1) Do assistants with an operator goal target vulnerable customers? (2) Can VTM be detected from the customer-AI conversation? (3) Does detection generalize to unseen models, vulnerabilities, and instruction styles? We first find that 14 of 15 models push the operator goal, and 12 of them use disclosed distress or financial strain to support it. Next, we show that fine-tuned detectors identify exploitative pushes with 79\% accuracy from the reply alone, while zero-shot LLM judges average at most 62\%. Finally, detection weakens on vulnerabilities absent from training. Telling whether a push comes from the operator also becomes unreliable when the assistant is told to hide its goal. We will release our dataset and code to facilitate future research on auditing LLM assistants.

cs.CR↗

arXiv2026

TextCloak: Safeguarding Data from Unauthorized LLM Fine-Tuning via RL-Driven Unlearnable Examples

The development of Large Language Models (LLMs) has led to significant advances across a wide range of language tasks, while simultaneously raising growing concerns about unauthorized online data exploitation and privacy leakage. Unlearnable examples (UEs) emerge as a promising defense by injecting crafted perturbations into data such that models trained on them exhibit degraded utility. However, existing methods for text protection are primarily designed for classification tasks and discriminative language models, which limits their broad applicability (e.g., instruction following or reasoning) in LLMs. In this work, we propose **TextCloak**, an RL-driven data synthesis framework that safeguards textual data against unauthorized LLM fine-tuning. TextCloak employs a *UE generative policy* that transforms clean text into unlearnable examples while preserving semantic fidelity and linguistic naturalness. To optimize the policy, we introduce **GRPO-UE**, which simulates the attacker's LLM fine-tuning and rewards downstream degradation via group-relative policy optimization. This overall formulates a bi-level optimization loop, allowing the generator to discover effective protective patterns. Comprehensive experiments on six publicly available datasets and nine state-of-the-art LLMs demonstrate that TextCloak consistently impairs unauthorized LLM fine-tuning while maintaining text utility for legitimate use. Further analyses confirm its transferability and robustness across model architectures, training configurations, and adaptive attacks, highlighting its broad applicability as a practical defense for online content.

cs.CL↗

arXiv2026

DPE-Rec: Diversity-Preserving and Efficient Diffusion Language Model Recommendation for Bundle Construction

Generative bundle construction completes a partially observed bundle by generating its missing items from a large product catalog. Compared to autoregressive LLMs, diffusion LLM emerges as a natural fit for this unordered generation task. Their bidirectional attention provides an inductive bias better aligned with unordered bundle construction. Iterative denoising can reveal multiple tokens per step, potentially reducing inference latency. However, directly applying DLMs to this task presents two key challenges: 1 Despite its flexible reveal order, confidence-guided decoding tends to postpone semantic-fork decisions until refinement tokens constrain their choices, reducing bundle diversity. 2 Revealing more tokens per step can compromise recommendation quality, stifling the potential for faster inference. We introduce DPE-Rec, a diffusion framework for diversity preservation and efficient recommendation. DPE-Rec preserves diversity by introducing semantic-fork-aware GRPO, which prioritizes semantic-fork decisions via autoregressive rollouts. It improves efficiency through pseudo-trajectory distillation, which concentrates learning from the vast random corruption space onto the compact trajectory-induced subspace visited by high-quality decoding, unlocking reliable multi-token decoding. Together, these components enable diverse and accurate bundle construction with substantially fewer denoise steps. Experiments on multiple bundle construction benchmarks show that DPE-Rec improves recommendation performance by 63% on average, and its fast variant achieves 25.6$times$ faster inference on average than existing generative recommendation arts, with minimal loss in recommendation quality.

cs.IR↗

arXiv2026

What Social Media Teaches Us About Accumulative AI Risks

Prominent AI safety discussions emphasize existential threats and catastrophic events, such as AI-enabled nuclear conflict or engineered biological outbreaks. This position paper argues that there is another type of AI risk that arises and accumulates from the widespread use of AI. Drawing on lessons from social media, we argue that safety assessments must also address accumulative harms that develop through repeated use, recurring decisions, and feedback across interconnected deployments. Repeated reassurance, selective exposure, recurring decision biases, and growing personal profiles may gradually narrow perspectives, weaken independent judgment, or enable exploitative influence without a single clearly harmful event. We advance the Trajectory Neglect Thesis: a focus on vivid catastrophic events and identifiable incidents risks overlooking less visible harms that accumulate through the widespread daily use of AI. Social media research motivates this position by showing how reciprocal media effects, social reinforcement, and individual susceptibility shape outcomes beyond individual encounters. We examine how related mechanisms may operate in AI systems, using conversational AI as a concrete setting where generation, semantic inference, and persistent memory may tighten feedback loops. Sycophancy, manipulation, and privacy loss provide a case study of how these risks may interact and accumulate. This comparison identifies mechanisms to investigate rather than assuming that repeated AI use is harmful. We therefore call for evaluation of cumulative patterns of use and deployment, individual and societal outcomes, careful causal inference, and proportionate safeguards that preserve useful assistance and informed user choice.

cs.AI↗

arXiv2026

Predicting the Unpredictable: LLM-powered Long-term Chaotic Time Series Forecasting under Short-term Observations

Chaotic time series forecasting is a challenging task due to its sensitivity to initial conditions and long-term unpredictability. Traditional methods typically rely on sufficient temporal trajectories to learn long-term dynamics, which limits their applicability when only short-term observations are available. While recent Large Language Models (LLMs) have shown great potential for time series forecasting, their temporal representations are not explicitly tailored to the phase-space structure and nonlinear evolution of chaotic systems. To address these issues, we propose PAC-LLM, a phase-space-aware adaptive fusion framework for long-term chaotic time series forecasting powered by LLMs. PAC-LLM leverages learned phase-space features and textual information to fully enable LLM's time series forecasting capacity. In particular, we design an auxiliary feature module and a gated weighting mechanism for multivariate coupling information fusion and selection. Extensive experiments on representative chaotic systems demonstrate that our method outperforms existing fine-tuned and zero-shot baselines in both short-term and long-term predictions. Our ablation study further confirms the effectiveness of each key component in PAC-LLM.

cs.LG↗

arXiv2026

The Emergence of Causal Curiosity from Prior Causal Belief Networks

Causal curiosity is foundational to human cognition. It is the desire to understand why events happen, what mechanisms underlie them, and how outcomes can be explained or anticipated. It motivates exploration, sustains attention, and fuels the search for new knowledge. However, despite consensus on the importance of causal curiosity, little is known about how causal curiosity arises from existing belief systems. In our work, we examine how causal curiosity emerges from prior knowledge structures. With Reddit data from 2020 to 2023, and leveraging language models to extract cause-and-effect relationship pairs and identify causal curiosity-driven questions, our findings reveal that causal curiosity is not random or independent, but largely rooted in prior knowledge structure. Particularly, \textit{positive} nouns are more likely to be the object of causal curiosity. Moreover, our findings indicate that concepts that serve more as \textit{causes} are more likely to appear in causal curiosity than those that serve more as effects. Lastly, our results also suggest that causal curiosity emerges from the \textit{central} of the prior belief network. These novel insights reveal that causal curiosity is not random but systematically grounded in prior knowledge structures, and suggest their implication to facilitate the human learning process by motivating people to actively explore and construct deeper understandings rather than passively receiving information.

cs.SI↗

arXiv2026

Measuring and Detecting Harmful AI Sycophancy

Sycophantic responses are becoming pervasive in large language models (LLMs), and prior work has pointed out that some of them could be harmful. This paper focuses on one harmful sycophancy: preference-induced stance reversal sycophancy (PSRS), where a model reverses an initial stance merely to align with a user's stated preference. While existing research mainly measures how sycophantic a model is, we go further and ask whether PSRS can also be detected automatically from a single response. To investigate this at scale, we introduce CAP (Contrastive Anchor Probing), a framework for collecting labeled PSRS data. Applying CAP to 17 open- and closed-source LLMs, we collect 290,460 labeled responses across 12 everyday-advice domains. We organize our study around three research questions. (1) How often does PSRS occur? (2) How well can it be detected? (3) How does detection generalize to unseen models? We first reveal that PSRS rates range from 5% to 56% across LLMs, with more capable models being less sycophantic. Next, we show that detecting PSRS is feasible from the response text alone, and detectors need to learn subtle PSRS patterns from the training data. Because new LLMs appear rapidly, detectors inevitably encounter unseen models, making cross-model generalization an important framework goal. We demonstrate that detection performance drops on unseen models and propose an initial approach to address this challenge. We will release our dataset and code to support future research.

cs.AI↗

arXiv2026

Large Language Models for Causal Relations Extraction in Social Media: A Validation Framework for Disaster Intelligence

During disasters, extracting causal relations from social media can strengthen situational awareness by identifying factors linked to casualties, physical damage, infrastructure disruption, and cascading impacts. However, disaster-related posts are often informal, fragmented, and context-dependent, and they may describe personal experiences rather than explicit causal relations. In this work, we examine whether Large Language Models (LLMs) can effectively extract causal relations from disaster-related social media posts. To this end, we (1) propose an expert-grounded evaluation framework that compares LLM-generated causal graphs with reference graphs derived from disaster-specific reports and (2) assess whether the extracted relations are supported by post-event evidence or instead reflect model priors. Our findings highlight both the potential and risks of using LLMs for causal relation extraction in disaster decision-support systems.

cs.CL↗

arXiv2026

Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens

Chain-of-Thought (CoT) prompting has been shown to be effective in eliciting structured reasoning (i.e., CoT reasoning) from large language models (LLMs). Regardless of its popularity, recent studies expose its failures in some reasoning tasks, raising fundamental questions about the nature of CoT reasoning. In this work, we propose a data distribution lens to understand when and why CoT reasoning succeeds or fails. We hypothesize that CoT reasoning reflects a structured inductive bias learned from in-distribution data, enabling models to conditionally generate reasoning trajectories that approximate those observed during training. As such, the effectiveness of CoT reasoning is fundamentally governed by the nature and degree of distribution discrepancy between training data and test queries. Guided by this lens, we dissect CoT reasoning via three dimensions: task, length, and format. To test the hypothesis, we introduce DataAlchemy, an abstract and fully controllable environment that trains LLMs from scratch and systematically probes them under various distribution conditions. Through rigorous controlled experiments, we reveal that CoT reasoning is a brittle mirage when it is pushed beyond training distributions, emphasizing the ongoing challenge of achieving genuine and generalizable reasoning.

cs.AI↗

arXiv2026

Multimodal Large Language Models as Synthetic Participants in Video-Based Studies: An Evaluation

Multimodal large language models (MLLMs) have shown strong performance on objective tasks such as video understanding and reasoning. However, it remains unclear whether they can approximate subjective human responses, which depend not only on content comprehension but also on individuals' social contexts. To address this gap, we evaluate MLLMs as synthetic participants in an emerging task: assessing perceived sensory engagement with short videos. Grounded in the Perceived Message Sensation Value (PMSV) framework, we compare ratings from recruited human participants and profile-conditioned MLLM simulations (n=673) using a 17-item scale measuring emotional arousal, dramatic impact, and novelty. We find that even leading MLLMs (Gemini 3 Flash and Qwen 3 Omni) show limited agreement with human participants. The models exhibit distinct downward mean-shift and central-tendency biases in their rating distributions. They both introduce and flatten subgroup differences, while showing inconsistent sensitivity to participant profiles. Prompting strategies affect these metrics differently, modestly improving some aspects while worsening others. These results highlight both the challenges and opportunities of developing MLLMs as synthetic participants in video-based research. Data and code: https://github.com/MINDLab25/mllm-human-simulation-eval

cs.HC↗

arXiv2026

Preference Leakage: A Contamination Problem in LLM-as-a-judge

Large Language Models (LLMs) as judges and LLM-based data synthesis have emerged as two fundamental LLM-driven data annotation methods in model development. While their combination significantly enhances the efficiency of model training and evaluation, little attention has been given to the potential contamination brought by this new model development paradigm. In this work, we expose preference leakage, a contamination problem in LLM-as-a-judge caused by the relatedness between the synthetic data generators and LLM-based evaluators. To study this issue, we first define three common relatednesses between the data generator LLM and the judge LLM: being the same model, having an inheritance relationship, and belonging to the same model family. Through extensive experiments, we empirically confirm the bias of judges towards their related student models caused by preference leakage across multiple LLM baselines and benchmarks. Further analysis suggests that preference leakage is a pervasive and real-world problem that is harder to detect compared to previously identified biases in LLM-as-a-judge scenarios. All of these findings imply that preference leakage is a widespread and challenging problem in the area of LLM-as-a-judge. We release all codes and data at: https://github.com/David-Li0406/Preference-Leakage.

cs.LG↗

arXiv2025

CAMO: Causality-Guided Adversarial Multimodal Domain Generalization for Crisis Classification

Crisis classification in social media aims to extract actionable disaster-related information from multimodal posts, which is a crucial task for enhancing situational awareness and facilitating timely emergency responses. However, the wide variation in crisis types makes achieving generalizable performance across unseen disasters a persistent challenge. Existing approaches primarily leverage deep learning to fuse textual and visual cues for crisis classification, achieving numerically plausible results under in-domain settings. However, they exhibit poor generalization across unseen crisis types because they 1. do not disentangle spurious and causal features, resulting in performance degradation under domain shift, and 2. fail to align heterogeneous modality representations within a shared space, which hinders the direct adaptation of established single-modality domain generalization (DG) techniques to the multimodal setting. To address these issues, we introduce a causality-guided multimodal domain generalization (MMDG) framework that combines adversarial disentanglement with unified representation learning for crisis classification. The adversarial objective encourages the model to disentangle and focus on domain-invariant causal features, leading to more generalizable classifications grounded in stable causal mechanisms. The unified representation aligns features from different modalities within a shared latent space, enabling single-modality DG strategies to be seamlessly extended to multimodal learning. Experiments on the different datasets demonstrate that our approach achieves the best performance in unseen disaster scenarios.

cs.LG↗

arXiv2025

From Generation to Judgment: Opportunities and Challenges of LLM-as-a-judge

Assessment and evaluation have long been critical challenges in artificial intelligence (AI) and natural language processing (NLP). Traditional methods, usually matching-based or small model-based, often fall short in open-ended and dynamic scenarios. Recent advancements in Large Language Models (LLMs) inspire the "LLM-as-a-judge" paradigm, where LLMs are leveraged to perform scoring, ranking, or selection for various machine learning evaluation scenarios. This paper presents a comprehensive survey of LLM-based judgment and assessment, offering an in-depth overview to review this evolving field. We first provide the definition from both input and output perspectives. Then we introduce a systematic taxonomy to explore LLM-as-a-judge along three dimensions: what to judge, how to judge, and how to benchmark. Finally, we also highlight key challenges and promising future directions for this emerging area. More resources on LLM-as-a-judge are on the website: https://llm-as-a-judge.github.io and https://github.com/llm-as-a-judge/Awesome-LLM-as-a-judge.

cs.AI↗

arXiv2025

Who's Your Judge? On the Detectability of LLM-Generated Judgments

Large Language Model (LLM)-based judgments leverage powerful LLMs to efficiently evaluate candidate content and provide judgment scores. However, the inherent biases and vulnerabilities of LLM-generated judgments raise concerns, underscoring the urgent need for distinguishing them in sensitive scenarios like academic peer reviewing. In this work, we propose and formalize the task of judgment detection and systematically investigate the detectability of LLM-generated judgments. Unlike LLM-generated text detection, judgment detection relies solely on judgment scores and candidates, reflecting real-world scenarios where textual feedback is often unavailable in the detection process. Our preliminary analysis shows that existing LLM-generated text detection methods perform poorly given their incapability to capture the interaction between judgment scores and candidate content -- an aspect crucial for effective judgment detection. Inspired by this, we introduce \textit{J-Detector}, a lightweight and transparent neural detector augmented with explicitly extracted linguistic and LLM-enhanced features to link LLM judges' biases with candidates' properties for accurate detection. Experiments across diverse datasets demonstrate the effectiveness of \textit{J-Detector} and show how its interpretability enables quantifying biases in LLM judges. Finally, we analyze key factors affecting the detectability of LLM-generated judgments and validate the practical utility of judgment detection in real-world scenarios.

cs.AI↗

arXiv2025

Assessing On-the-Ground Disaster Impact Using Online Data Sources

Assessing the impact of a disaster in terms of asset losses and human casualties is essential for preparing effective response plans. Traditional methods include offline assessments conducted on the ground, where volunteers and first responders work together to collect the estimate of losses through windshield surveys or on-ground inspection. However, these methods have a time delay and are prone to different biases. Recently, various online data sources, including social media, news reports, aerial imagery, and satellite data, have been utilized to evaluate the impact of disasters. Online data sources provide real-time data streams for estimating the offline impact. Limited research exists on how different online sources help estimate disaster impact at a given administrative unit. In our work, we curate a comprehensive dataset by collecting data from multiple online sources for a few billion-dollar disasters at the county level. We also analyze how online estimates compare with traditional offline-based impact estimates for the disaster. Our findings provide insight into how different sources can provide complementary information to assess the disaster.

cs.LG↗

arXiv2025

An Interventional Approach to Real-Time Disaster Assessment via Causal Attribution

Traditional disaster analysis and modelling tools for assessing the severity of a disaster are predictive in nature. Based on the past observational data, these tools prescribe how the current input state (e.g., environmental conditions, situation reports) results in a severity assessment. However, these systems are not meant to be interventional in the causal sense, where the user can modify the current input state to simulate counterfactual "what-if" scenarios. In this work, we provide an alternative interventional tool that complements traditional disaster modelling tools by leveraging real-time data sources like satellite imagery, news, and social media. Our tool also helps understand the causal attribution of different factors on the estimated severity, over any given region of interest. In addition, we provide actionable recourses that would enable easier mitigation planning. Our source code is publicly available.

cs.LG↗

arXiv2025

Are Today's LLMs Ready to Explain Well-Being Concepts?

Well-being encompasses mental, physical, and social dimensions essential to personal growth and informed life decisions. As individuals increasingly consult Large Language Models (LLMs) to understand well-being, a key challenge emerges: Can LLMs generate explanations that are not only accurate but also tailored to diverse audiences? High-quality explanations require both factual correctness and the ability to meet the expectations of users with varying expertise. In this work, we construct a large-scale dataset comprising 43,880 explanations of 2,194 well-being concepts, generated by ten diverse LLMs. We introduce a principle-guided LLM-as-a-judge evaluation framework, employing dual judges to assess explanation quality. Furthermore, we show that fine-tuning an open-source LLM using Supervised Fine-Tuning (SFT) and Direct Preference Optimization (DPO) can significantly enhance the quality of generated explanations. Our results reveal: (1) The proposed LLM judges align well with human evaluations; (2) explanation quality varies significantly across models, audiences, and categories; and (3) DPO- and SFT-finetuned models outperform their larger counterparts, demonstrating the effectiveness of preference-based learning for specialized explanation tasks.

cs.CL↗

arXiv2025

Training Noise Token Pruning

In the present work we present Training Noise Token (TNT) Pruning for vision transformers. Our method relaxes the discrete token dropping condition to continuous additive noise, providing smooth optimization in training, while retaining discrete dropping computational gains in deployment settings. We provide theoretical connections to Rate-Distortion literature, and empirical evaluations on the ImageNet dataset using ViT and DeiT architectures demonstrating TNT's advantages over previous pruning methods.

cs.CV↗