Evaluating Language Models for Harmful Manipulation

Interest in the concept of AI-driven harmful manipulation is growing, yet current approaches to evaluating it are limited. This paper introduces a framework for evaluating harmful AI manipulation via context-specific human-AI interaction studies. We illustrate the utility of this framework by assessing an AI model with 10,101 participants spanning interactions in three AI use domains (public policy, finance, and health) and three locales (US, UK, and India). Overall, we find that that the tested model can produce manipulative behaviours when prompted to do so and, in experimental settings, is able to induce belief and behaviour changes in study participants. We further find that context matters: AI manipulation differs between domains, suggesting that it needs to be evaluated in the high-stakes context(s) in which an AI system is likely to be used. We also identify significant differences across our tested geographies, suggesting that AI manipulation results from one geographic region may not generalise to others. Finally, we find that the frequency of manipulative behaviours (propensity) of an AI model is not consistently predictive of the likelihood of manipulative success (efficacy), underscoring the importance of studying these dimensions separately. To facilitate adoption of our evaluation framework, we detail our testing protocols and make relevant materials publicly available. We conclude by discussing open challenges in evaluating harmful manipulation by AI models.

A Minimum-Energy Control Approach for Redundant Mobile Manipulators in Physical Human-Robot Interaction Applications

Research on mobile manipulation systems that physically interact with humans has expanded rapidly in recent years, opening the way to tasks which could not be performed using fixed-base manipulators. Within this context, developing suitable control methodologies is essential since mobile manipulators introduce additional degrees of freedom, making the design of control approaches more challenging and more prone to performance optimization. This paper proposes a control approach for a mobile manipulator, composed of a mobile base equipped with a robotic arm mounted on the top, with the objective of minimizing the overall kinetic energy stored in the whole-body mobile manipulator in physical human-robot interaction applications. The approach is experimentally tested with reference to a peg-in-hole task, and the results demonstrate that the proposed approach reduces the overall kinetic energy stored in the whole-body robotic system and improves the system performance compared with the benchmark method.

STN-GPR: A Singularity Tensor Network Framework for Efficient Option Pricing

We develop a tensor-network surrogate for option pricing, targeting large-scale portfolio revaluation problems arising in market risk management (e.g., VaR and Expected Shortfall computations). The method involves representing high-dimensional price surfaces in tensor-train (TT) form using TT-cross approximation, constructing the surrogate directly from black-box price evaluations without materializing the full training tensor. For inference, we use a Laplacian kernel and derive TT representations of the kernel matrix and its closed-form inverse in the noise-free setting, enabling TT-based Gaussian process regression without dense matrix factorization or iterative linear solves. We found that hyperparameter optimization consistently favors a large kernel length-scale and show that in this regime the GPR predictor reduces to multilinear interpolation for off-grid inputs; we also derive a low-rank TT representation for this limit. We evaluate the approach on five-asset basket options over an eight dimensional parameter space (asset spot levels, strike, interest rate, and time to maturity). For European geometric basket puts, the tensor surrogate achieves lower test error at shorter training times than standard GPR by scaling to substantially larger effective training sets. For American arithmetic basket puts trained on LSMC data, the surrogate exhibits more favorable scaling with training-set size while providing millisecond-level evaluation per query, with overall runtime dominated by data generation.

Clinical named entity recognition in the Portuguese language: a benchmark of modern BERT models and LLMs

Clinical notes contain valuable unstructured information. Named entity recognition (NER) enables the automatic extraction of medical concepts; however, benchmarks for Portuguese remain scarce. In this study, we aimed to evaluate BERT-based models and large language models (LLMs) for clinical NER in Portuguese and to test strategies for addressing multilabel imbalance. We compared BioBERTpt, BERTimbau, ModernBERT, and mmBERT with LLMs such as GPT-5 and Gemini-2.5, using the public SemClinBr corpus and a private breast cancer dataset. Models were trained under identical conditions and evaluated using precision, recall, and F1-score. Iterative stratification, weighted loss, and oversampling were explored to mitigate class imbalance. The mmBERT-base model achieved the best performance (micro F1 = 0.76), outperforming all other models. Iterative stratification improved class balance and overall performance. Multilingual BERT models, particularly mmBERT, perform strongly for Portuguese clinical NER and can run locally with limited computational resources. Balanced data-splitting strategies further enhance performance.

Semi-structured multi-state delinquency model for mortgage default

We propose a semi-structured discrete-time multi-state model to analyse mortgage delinquency transitions. This model combines an easy-to-understand structured additive predictor, which includes linear effects and smooth functions of time and covariates, with a flexible neural network component that captures complex nonlinearities and higher-order interactions. To ensure identifiability when covariates are present in both components, we orthogonalise the unstructured part relative to the structured design. For discrete-time competing transitions, we derive exact transformations that map binary logistic models to valid competing transition probabilities, avoiding the need for continuous-time approximations. In simulations, our framework effectively recovers structured baseline and covariate effects while using the neural component to detect interaction patterns. We demonstrate the method using the Freddie Mac Single-Family Loan-Level Dataset, employing an out-of-time test design. Compared with a structured generalised additive benchmark, the semi-structured model provides modest but consistent gains in discrimination across the earliest prediction spans, while maintaining similar Brier scores. Adding macroeconomic indicators provides limited incremental benefit in this out-of-time evaluation and does not materially change the estimated borrower-, loan-, or duration-driven effects. Overall, semi-structured multi-state modelling offers a practical compromise between transparent effect estimates and flexible pattern learning, with potential applications beyond credit-transition forecasting.

THEMIS: Towards Holistic Evaluation of MLLMs for Scientific Paper Fraud Forensics

We present THEMIS, a novel multi-task benchmark designed to comprehensively evaluate multimodal large language models (MLLMs) on visual fraud reasoning within real-world academic scenarios. Compared to existing benchmarks, THEMIS introduces three major advances. (1) Real-World Scenarios and Complexity: Our benchmark comprises over 4,000 questions spanning seven scenarios, derived from authentic retracted-paper cases and carefully curated multimodal synthetic data. With 60.47% complex-texture images, THEMIS bridges the critical gap between existing benchmarks and the complexity of real-world academic fraud. (2) Fraud-Type Diversity and Granularity: THEMIS systematically covers five challenging fraud types and introduces 16 fine-grained manipulation operations. On average, each sample undergoes multiple stacked manipulation operations, with the diversity and difficulty of these manipulations demanding a high level of visual fraud reasoning from the models. (3) Multi-Dimensional Capability Evaluation: We establish a mapping from fraud types to five core visual fraud reasoning capabilities, thereby enabling an evaluation that reveals the distinct strengths and specific weaknesses of different models across these core capabilities. Experiments on 16 leading MLLMs show that even the best-performing model, GPT-5, achieves an overall performance of only 56.15%, demonstrating that our benchmark presents a stringent test. We expect THEMIS to advance the development of MLLMs for complex, real-world fraud reasoning tasks.

Adapting Frozen Mono-modal Backbones for Multi-modal Registration via Contrast-Agnostic Instance Optimization

Deformable image registration remains a central challenge in medical image analysis, particularly under multi-modal scenarios where intensity distributions vary significantly across scans. While deep learning methods provide efficient feed-forward predictions, they often fail to generalize robustly under distribution shifts at test time. A straightforward remedy is full network fine-tuning, yet for modern architectures such as Transformers or deep U-Nets, this adaptation is prohibitively expensive in both memory and runtime when operating in 3D. Meanwhile, the naive fine-tuning struggles more with potential degradation in performance in the existence of drastic domain shifts. In this work, we propose a registration framework that integrates a frozen pretrained \textbf{mono-modal} registration model with a lightweight adaptation pipeline for \textbf{multi-modal} image registration. Specifically, we employ style transfer based on contrast-agnostic representation generation and refinement modules to bridge modality and domain gaps with instance optimization at test time. This design is orthogonal to the choice of backbone mono-modal model, thus avoids the computational burden of full fine-tuning while retaining the flexibility to adapt to unseen domains. We evaluate our approach on the Learn2Reg 2025 LUMIR validation set and observe consistent improvements over the pretrained state-of-the-art mono-modal backbone. In particular, the method ranks second on the multi-modal subset, third on the out-of-domain subset, and achieves fourth place overall in Dice score. These results demonstrate that combining frozen mono-modal models with modality adaptation and lightweight instance optimization offers an effective and practical pathway toward robust multi-modal registration.

Meta-Adaptive Beam Search Planning for Transformer-Based Reinforcement Learning Control of UAVs with Overhead Manipulators under Flight Disturbances

Drones equipped with overhead manipulators offer unique capabilities for inspection, maintenance, and contact-based interaction. However, the motion of the drone and its manipulator is tightly linked, and even small attitude changes caused by wind or control imperfections shift the end-effector away from its intended path. This coupling makes reliable tracking difficult and also limits the direct use of learning-based arm controllers that were originally designed for fixed-base robots. These effects appear consistently in our tests whenever the UAV body experiences drift or rapid attitude corrections. To address this behavior, we develop a reinforcement-learning (RL) framework with a transformer-based double deep Q learning (DDQN), with the core idea of using an adaptive beam-search planner that applies a short-horizon beam search over candidate control sequences using the learned critic as the forward estimator. This allows the controller to anticipate the end-effector's motion through simulated rollouts rather than executing those actions directly on the actual model, realizing a software-in-the-loop (SITL) approach. The lookahead relies on value estimates from a Transformer critic that processes short sequences of states, while a DDQN backbone provides the one-step targets needed to keep the learning process stable. Evaluated on a 3-DoF aerial manipulator under identical training conditions, the proposed meta-adaptive planner shows the strongest overall performance with a 10.2% reward increase, a substantial reduction in mean tracking error (from about 6% to 3%), and a 29.6% improvement in the combined reward-error metric relative to the DDQN baseline. Our method exhibits elevated stability in tracking target tip trajectory (by maintaining 5 cm tracking error) when the drone base exhibits drifts due to external disturbances, as opposed to the fixed-beam and Transformer-only variants.

Separate Before You Compress: The WWHO Tokenization Architecture

Current Large Language Models (LLMs) mostly use BPE (Byte Pair Encoding) based tokenizers, which are very effective for simple structured Latin scripts such as English. However, standard BPE tokenizers struggle to process complex Abugida scripts due to their structural complexity. The problem is that these tokenizers break complex conjuncts, which are multi-codepoint grapheme clusters, into meaningless sub-character units. This degrades the LLM's reasoning efficiency by forcing it to learn basic orthographic structures at inference time and raises inference costs, resulting in a significant "Token Tax" for the Global South. We propose a new three-layer architecture, the WWHO (Where-What-How Often), and an algorithm named SGPE (Syllable-aware Grapheme Pair Encoding) that separates the linguistic rules of the script from the statistical compression process while enabling seamless multilingual tokenization. Using Sinhala and Devanagari (Hindi/Sanskrit) as highly complex Abugida scripts, we trained WWHO on a cleaned 30-million-sentence dataset and evaluated on a 1,499,950-sentence test set. For Sinhala, SGPE achieves a Token to Word Ratio (TWR) of 1.274 with 4.83 characters per token, representing a 61.7 percent reduction in tokens compared to OpenAI's o200k base. For Hindi, it achieves a TWR of 1.181 (27.0 percent reduction vs o200k). On the mixed-script (Sinhala, Devanagari, and English) dataset, SGPE achieves an overall TWR of 1.240, representing token reductions of 36.7 percent, 39.6 percent, and 60.2 percent relative to o200k base, Llama 4 Scout, and DeepSeek V3, respectively. This effectively extends the usable context window by up to 4.38 times for these Abugida languages while ensuring a Linguistic Zero-Breakage Guarantee, which ensures that no valid syllable is ever split across multiple tokens.

Can Small Models Reason About Legal Documents? A Comparative Study

Large language models show promise for legal applications, but deploying frontier models raises concerns about cost, latency, and data privacy. We evaluate whether sub-10B parameter models can serve as practical alternatives by testing nine models across three legal benchmarks (ContractNLI, CaseHOLD, and ECtHR) using five prompting strategies (direct, chain-of-thought, few-shot, BM25 RAG, and dense RAG). Across 405 experiments with three random seeds per configuration, we find that a Mixture-of-Experts model activating only 3B parameters matches GPT-4o-mini in mean accuracy while surpassing it on legal holding identification, and that architecture and training quality matter more than raw parameter count. Our largest model (9B parameters) performs worst overall. Chain-of-thought prompting proves sharply task-dependent, improving contract entailment but degrading multiple-choice legal reasoning, while few-shot prompting emerges as the most consistently effective strategy. Comparing BM25 and dense retrieval for RAG, we find near-identical results, suggesting the bottleneck lies in the language model's utilization of retrieved context rather than retrieval quality. All experiments were conducted via cloud inference APIs at a total cost of $62, demonstrating that rigorous LLM evaluation is accessible without dedicated GPU infrastructure.