We introduce the concept of "Legal Zero-Days" as a novel risk vector for advanced AI systems. Legal Zero-Days are previously undiscovered vulnerabilities in legal frameworks that, when exploited, can cause immediate and significant societal disruption without requiring litigation or other processes before impact. We present a risk model for identifying and evaluating these vulnerabilities, demonstrating their potential to bypass safeguards or impede government responses to AI incidents. Using the 2017 Australian dual citizenship crisis as a case study, we illustrate how seemingly minor legal oversights can lead to large-scale governance disruption. We develop a methodology for creating "legal puzzles" as evaluation instruments for assessing AI systems' capabilities to discover such vulnerabilities. Our findings suggest that while current AI models may not reliably find impactful Legal Zero-Days, future systems may develop this capability, presenting both risks and opportunities for improving legal robustness. This work contributes to the broader effort to identify and mitigate previously unrecognized risks from frontier AI systems.

By virtue of its great utility in solving real-world problems, optimization modeling has been widely employed for optimal decision-making across various sectors, but it requires substantial expertise from operations research professionals. With the advent of large language models (LLMs), new opportunities have emerged to automate the procedure of mathematical modeling. This survey presents a comprehensive and timely review of recent advancements that cover the entire technical stack, including data synthesis and fine-tuning for the base model, inference frameworks, benchmark datasets, and performance evaluation. In addition, we conducted an in-depth analysis on the quality of benchmark datasets, which was found to have a surprisingly high error rate. We cleaned the datasets and constructed a new leaderboard with fair performance evaluation in terms of base LLM model and datasets. We also build an online portal that integrates resources of cleaned datasets, code and paper repository to benefit the community. Finally, we identify limitations in current methodologies and outline future research opportunities.

Reinforcement learning in large reasoning models enables learning from feedback on their outputs, making it particularly valuable in scenarios where fine-tuning data is limited. However, its application in multi-modal human activity recognition (HAR) domains remains largely underexplored. Our work extends reinforcement learning to the human activity recognition domain with multimodal large language models. By incorporating visual reinforcement learning in the training process, the model's generalization ability on few-shot recognition can be greatly improved. Additionally, visual reinforcement learning can enhance the model's reasoning ability and enable explainable analysis in the inference stage. We name our few-shot human activity recognition method with visual reinforcement learning FAVOR. Specifically, our approach first utilizes a multimodal large language model (MLLM) to generate multiple candidate responses for the human activity image, each containing reasoning traces and final answers. These responses are then evaluated using reward functions, and the MLLM model is subsequently optimized using the Group Relative Policy Optimization (GRPO) algorithm. In this way, the MLLM model can be adapted to human activity recognition with only a few samples. Extensive experiments on four human activity recognition datasets and five different settings demonstrate the superiority of the proposed method.

Modern vision-language models (VLMs) often fail at cultural competency evaluations and benchmarks. Given the diversity of applications built upon VLMs, there is renewed interest in understanding how they encode cultural nuances. While individual aspects of this problem have been studied, we still lack a comprehensive framework for systematically identifying and annotating the nuanced cultural dimensions present in images for VLMs. This position paper argues that foundational methodologies from visual culture studies (cultural studies, semiotics, and visual studies) are necessary for cultural analysis of images. Building upon this review, we propose a set of five frameworks, corresponding to cultural dimensions, that must be considered for a more complete analysis of the cultural competencies of VLMs.

We investigate the potential of large language models (LLMs) to serve as efficient simulators for agentic search tasks in reinforcement learning (RL), thereby reducing dependence on costly interactions with external search engines. To this end, we first quantify the intrinsic search capability of LLMs via structured prompting and repeated sampling, which we term Self-Search. Our results reveal that LLMs exhibit strong scaling behavior with respect to the inference budget, achieving high pass@k on question-answering benchmarks, including the challenging BrowseComp task. Building on these observations, we introduce Self-Search RL (SSRL), which enhances LLMs' Self-Search capability through format-based and rule-based rewards. SSRL enables models to iteratively refine their knowledge utilization internally, without requiring access to external tools. Empirical evaluations demonstrate that SSRL-trained policy models provide a cost-effective and stable environment for search-driven RL training, reducing reliance on external search engines and facilitating robust sim-to-real transfer. We draw the following conclusions: 1) LLMs possess world knowledge that can be effectively elicited to achieve high performance; 2) SSRL demonstrates the potential of leveraging internal knowledge to reduce hallucination; 3) SSRL-trained models integrate seamlessly with external search engines without additional effort. Our findings highlight the potential of LLMs to support more scalable RL agent training.

Accurate differentiation of brain tumor types on magnetic resonance imaging (MRI) is critical for guiding treatment planning in neuro-oncology. Recent advances in large language models (LLMs) have enabled visual question answering (VQA) approaches that integrate image interpretation with natural language reasoning. In this study, we evaluated GPT-4o, GPT-5-nano, GPT-5-mini, and GPT-5 on a curated brain tumor VQA benchmark derived from 3 Brain Tumor Segmentation (BraTS) datasets - glioblastoma (GLI), meningioma (MEN), and brain metastases (MET). Each case included multi-sequence MRI triplanar mosaics and structured clinical features transformed into standardized VQA items. Models were assessed in a zero-shot chain-of-thought setting for accuracy on both visual and reasoning tasks. Results showed that GPT-5-mini achieved the highest macro-average accuracy (44.19%), followed by GPT-5 (43.71%), GPT-4o (41.49%), and GPT-5-nano (35.85%). Performance varied by tumor subtype, with no single model dominating across all cohorts. These findings suggest that GPT-5 family models can achieve moderate accuracy in structured neuro-oncological VQA tasks, but not at a level acceptable for clinical use.

Recent advancements in machine learning have spurred growing interests in automated interpreting quality assessment. Nevertheless, existing research suffers from insufficient examination of language use quality, unsatisfactory modeling effectiveness due to data scarcity and imbalance, and a lack of efforts to explain model predictions. To address these gaps, we propose a multi-dimensional modeling framework that integrates feature engineering, data augmentation, and explainable machine learning. This approach prioritizes explainability over ``black box'' predictions by utilizing only construct-relevant, transparent features and conducting Shapley Value (SHAP) analysis. Our results demonstrate strong predictive performance on a novel English-Chinese consecutive interpreting dataset, identifying BLEURT and CometKiwi scores to be the strongest predictive features for fidelity, pause-related features for fluency, and Chinese-specific phraseological diversity metrics for language use. Overall, by placing particular emphasis on explainability, we present a scalable, reliable, and transparent alternative to traditional human evaluation, facilitating the provision of detailed diagnostic feedback for learners and supporting self-regulated learning advantages not afforded by automated scores in isolation.

Large Language Models (LLMs) show potential as sequential decision-making agents, but their application is often limited due to a reliance on large, computationally expensive models. This creates a need to improve smaller models, yet existing post-training methods are designed for single-turn interactions and cannot handle credit assignment in multi-step agentic tasks. To address this, we introduce Multi-Step Group-Relative Policy Optimization (MS-GRPO), a new algorithm for post-training LLM agents, grounded in formal Text-Mediated Stochastic Game (TSMG) and Language-Agent Policy (LAP) frameworks. For credit assignment, MS-GRPO attributes the entire cumulative episode reward to each individual episode step. We supplement this algorithm with a novel absolute-advantage-weighted episode sampling strategy that we show improves training performance. We evaluate our approach by post-training a 3-billion parameter model on Snake and Frozen Lake. Our experiments demonstrate that the method is effective in improving decision-making performance: our post-trained 3B parameter model outperforms a 72B parameter baseline by 50% on the Frozen Lake task. This work demonstrates that targeted post-training is a practical and efficient alternative to relying on model scale for creating sequential decision-making agents using LLMs.

Data minimization (DM) describes the principle of collecting only the data strictly necessary for a given task. It is a foundational principle across major data protection regulations like GDPR and CPRA. Violations of this principle have substantial real-world consequences, with regulatory actions resulting in fines reaching hundreds of millions of dollars. Notably, the relevance of data minimization is particularly pronounced in machine learning (ML) applications, which typically rely on large datasets, resulting in an emerging research area known as Data Minimization in Machine Learning (DMML). At the same time, existing work on other ML privacy and security topics often addresses concerns relevant to DMML without explicitly acknowledging the connection. This disconnect leads to confusion among practitioners, complicating their efforts to implement DM principles and interpret the terminology, metrics, and evaluation criteria used across different research communities. To address this gap, our work introduces a comprehensive framework for DMML, including a unified data pipeline, adversaries, and points of minimization. This framework allows us to systematically review the literature on data minimization and \emph{DM-adjacent} methodologies, for the first time presenting a structured overview designed to help practitioners and researchers effectively apply DM principles. Our work facilitates a unified DM-centric understanding and broader adoption of data minimization strategies in AI/ML.

-- Subjective self-disclosure is an important feature of human social interaction. While much has been done in the social and behavioural literature to characterise the features and consequences of subjective self-disclosure, little work has been done thus far to develop computational systems that are able to accurately model it. Even less work has been done that attempts to model specifically how human interactants self-disclose with robotic partners. It is becoming more pressing as we require social robots to work in conjunction with and establish relationships with humans in various social settings. In this paper, our aim is to develop a custom multimodal attention network based on models from the emotion recognition literature, training this model on a large self-collected self-disclosure video corpus, and constructing a new loss function, the scale preserving cross entropy loss, that improves upon both classification and regression versions of this problem. Our results show that the best performing model, trained with our novel loss function, achieves an F1 score of 0.83, an improvement of 0.48 from the best baseline model. This result makes significant headway in the aim of allowing social robots to pick up on an interaction partner's self-disclosures, an ability that will be essential in social robots with social cognition. Self-disclosure is the sharing of one's thoughts, feelings, or personal information during a social interaction [1]. It is an import facet of human social behaviour and can contribute to many aspects of our lives.