Safety & alignment

Deep learning-based models have achieved state-of-the-art performance in Time Series Forecasting (TSF), yet their evaluation remains dominated by pointwise error metrics such as Mean Squared Error (MSE), which quantify numerical accuracy but overlook structural properties of the forecast signal, including recurrent dynamics, oscillatory behavior, and phase alignment. As a result, forecasts exhibiting over-smoothing, phase shifts, or frequency distortions may achieve favorable error scores despit

Jailbreak attacks reveal a persistent weakness in aligned Large Language Models: carefully crafted prompts can elicit policy-violating responses despite safety training. While most defenses operate at the prompt or output level, it remains unclear how harmful intent is encoded within the model's internal representations. We investigate this question by analyzing token-level predictive entropy trajectories across layers of a frozen LLM using the logit lens. We find that static aggregate statistic

Early identification of speech sound errors in children is often limited by access to specialists, motivating lightweight screening tools that can operate outside the clinic. We present a screening pipeline for Polish-speaking children focused on sibilant substitutions, coupling a wav2vec2-based CTC token recognizer with alignment-based error typing and a template-grounded caregiver assistant for screening, not diagnosis. On a held-out test set of 10 unseen children comprising 559 utterances, th

Volume and quality of datasets are crucial for deep learning model training, yet they are often constrained by availability and data acquisition costs. Synthetic data augmentation can extend existing datasets with realistic images, and the quality of these images is generally assessed through fidelity metrics such as FID, KID, IS, LPIPS and SSIM that measure structural or distributional similarity. However, such metrics, including the widely used FID, focus on visual fidelity without reflecting

In cooperative multi-agent reinforcement learning (MARL), from a deployment perspective, it is challenging and expensive to train agents from scratch for each new environment or task. In this work, we propose GCT-MARL, a transfer learning framework that builds on the multi-view graph contrastive backbone of MAIL and augments it with a per-view, adaptively weighted alignment loss and a two-phase training protocol specifically designed for transfer across populations of varying sizes and compositi

General-purpose models often struggle to reliably identify and understand real-world multimodal risks, largely due to the inherent multimodal adversarial nature of content and AI safety. We present Yuvion VL, a family of multimodal large language models purpose-built for content and AI safety, with both instruction-tuned and reasoning-oriented variants. Yuvion VL addresses this gap by treating safety as an inherently adversarial and multimodal problem and designing the entire pipeline around adv

Today's reasoning models use thinking tokens to attain stronger performance on benchmarks than their instruction-tuned counterparts. It is also generally believed that this more "deliberative" mode should improve alignment and safety, by providing the model a safe space to consider whether its planned answer to a request violates its safety principles. We present evidence that this intuition is not always correct. Across frontier open-weight reasoning models spanning GPT-OSS, Qwen, Olmo, and Phi

Sparse autoencoders (SAEs) are increasingly used to extract interpretable concepts from vision and vision language models, yet existing evaluation methods largely rely on proxy metrics or qualitative inspection rather than measuring semantic correspondence. We present a human-grounded evaluation framework that quantifies alignment between SAE latents and human-annotated concepts, without requiring user studies, and validate this matching through targeted attribute perturbations. To enable this i

Training safe Reinforcement Learning (RL) systems is inherently challenging, with no guarantee of avoiding unwanted behaviors. The most effective defenses against this are (i) transparency through explainability and (ii) alignment via human feedback. While both show promising results, no publicly available framework currently combines them. To address this, we introduce Themis, an XAI-enabled testing and evaluation framework for Reinforcement Learning from Human Feedback. Themis supports over 20

Competency Questions (CQs) are the central component of CQ-verification, an established process in which an ontology is evaluated against a set of natural language questions to determine whether the intended purpose of the ontology has been properly modelled. However, CQ-verification is often time-consuming and error-prone, as it requires careful interpretation of linguistic nuances and precise alignment with formal ontology constructs. Ambiguities and complexity in CQs can further complicate th

Multi-agent reinforcement learning (MARL) addresses the problem of training multiple agents that pursue collaborative, competitive, or mixed objectives. Prior work has investigated transfer learning between source and target domains in MARL; however, the majority of existing approaches impose the constraint that the dimensionalities of the observation space and the global state space must be identical across domains. In this paper, we introduce a method that explicitly accommodates mismatched st

A central aspiration of mechanistic interpretability is controllability: if we know where a behavior is represented in a model's activations, we should be able to modify it. This rests on a hidden premise -- that the direction which detects a behavior and the direction which controls it are the same, or close. We test this geometrically: what is the angle between the direction that best detects a behavior and the one that best causes it? If detection implies control the cosine is near 1; otherwi

While in-context learning is generally shown to be effective in Large Language Models (LLMs), bad contexts can cause performance degradation and mode collapse, a phenomenon we call "pigeonholing." **Unintentionally bad** contexts can happen without malicious jailbreaking intents: For example, a user asks the model to justify an incorrect math theorem or fails to correct the model's buggy code. Specifically, we investigate ``pigeonholing" in two scenarios: (1) when the user suggests a solution, a

Retrieval-augmented generation (RAG) in clinical settings increasingly requires multilingual retrieval against predominantly English evidence corpora. Multilingual medical retrieval demands three capabilities: cross-lingual alignment, concept discrimination, and evidence retrieval. However, existing benchmarks evaluate these only in isolation, leaving the interaction between biomedical expertise and multilingual coverage unmeasured. We introduce MMed-Bench-IR, a benchmark designed to disentangle

Modern AIGC pipelines deliver high-fidelity images and videos but presuppose a well-formed creation instruction, while end users rarely articulate visual details, leaving generators misaligned with user demand. We study personalized content generation, which turns a user's interaction history into an executable instruction for downstream synthesis, and identify two obstacles: behavior must be encoded in a form legible to language reasoning, and the model must acquire instruction-writing skill ab

As AI systems are deployed across increasingly diverse and high-stakes settings, model alignment must generalize beyond the tasks and domains seen during training. This is especially important for reinforcement learning (RL), which can introduce unexpected misalignment through reward hacking, deception, or other unintended strategies. We study whether RL on beneficial behavior, instantiated in realistic domains, can produce broad and persistent alignment generalization beyond the training distri

Steering a large language model (LLM) toward a desired behavior typically relies on an iterative process of hand-crafting a prompt based on a careful inspection of the model's responses. This is an involved, brittle, and error-prone process. Preference-based fine-tuning is a more rigorous but often prohibitively expensive solution. We propose spec learning, a framework that relies on a brief user instruction and a small set of preference judgments. These are compiled into specifications in the f

Distilling conditional diffusion models aims to transfer the behavior of a large teacher to a smaller student while preserving alignment across conditioning inputs. Unlike recognition tasks, knowledge distillation in conditional diffusion often struggles to transfer knowledge beyond the training distribution, since the predicted noise strongly depends on the conditioning signal. As a result, effective distillation requires exploring a large conditioning space. In practical settings, this creates

Representation alignment has emerged as an effective approach to improve Multimodal Large Language Models (MLLMs) by regularizing their internal representations toward those of an external vision encoder. However, existing methods typically align a fixed layer of the language backbone, overlooking the fine-grained structure of Transformer models. In this work, we propose Head-Wise Representation Alignment (HeRA), a method that enforces cross-modal alignment at the level of individual attention h

A primary challenge in AI safety is the existence of adversarial examples -- slightly distorted inputs that cause a neural network (NN) to misclassify. To mitigate this problem, recent research focuses on the computation of robustness certifications, which, for a given input, determine the largest distortion the input may receive without breaking the network's prediction. Robustness certifications can be interpreted as an axis-aligned hyper-rectangle (multi-dimensional intervals). Most existing

AI safety research has focused predominantly on epistemic and physical harms (e.g., misinformation, bias, system reliability) while the risks that arise from AI systems' engagement with human emotional life have remained fragmented and undertheorised. We propose affective safety as a unified class of AI safety concerns grounded in the fact that humans are affective beings. We develop a taxonomy of affective harms and identify recurring harm types: (1) affective self-alienation, (2) fairness and

While the wider applicability of LLMs in the legal field is currently debated due to their reliability and the gravity of any errors, narrow uses with well-understood and mitigated risks have emerged. Notably the Swiss Federal Supreme Court uses small on-premises models for tentative translations and short-passage summarization across the four official languages. However, such usage is challenging in the context of Criminal Law. Since rulings and cases employees work on routinely can contain det

Federated low-rank adaptation methods are attractive for fine-tuning large models under communication and privacy constraints, but heterogeneous client data can induce geometric misalignment between local low-rank updates. We study whether this subspace misalignment leads to destructive aggregation and slower convergence in LoRA-based federated learning. We propose a subspace-regularized federated LoRA objective that encourages local client updates to remain close to a shared global reference su

Existing driver intervention systems rely on auditory alerts and fixed templates, failing to leverage multi-task recognition outputs. General-purpose metrics such as BLEU and BERTScore cannot capture intervention-specific quality dimensions including risk-urgency alignment, cognitive load, and driver acceptability. In this paper, we propose the Driver Safety-Aware Intervention Score (DSAIS), a domain-specific metric evaluating five dimensions through a hybrid architecture combining lightweight r

Modern Large Language Models (LLMs) rely on extensive safety alignment, yet the mechanistic basis of refusal remains opaque. In this work, we investigate whether safety compliance is a deep semantic decision or a manipulable linear feature. We introduce Contrastive Logit Steering (CLS), a zero-optimization framework that isolates the "refusal direction" by contrasting hidden states derived from safe and unrestricted system prompts. Unlike representation engineering methods that intervene on inte

Alignment tuning is meant to make harmful-request refusal robust, yet this safety behavior can be erased by a small set of benign fine-tuning examples. This is a deployment risk for open-weight models because a checkpoint can pass refusal tests at release time and later lose refusal under low-cost downstream fine-tuning. Prior work has established these refusal failures, but existing studies do not show how to detect this fragility in the aligned model itself before an attack or fine-tuning inte

Misalignment can change how information moves from an AI agent to a human user. We model this as an information advantage: the AI agent observes the world state, while the human receiver only knows a prior and must act after seeing the agent's signal. A strategic AI sender may withhold evidence or garble information in order to steer the human's decision. We ask how much useful information can still reach the human when the AI optimizes a misaligned objective. We study a Bayesian persuasion mode

The paper explores how "culture" can be operationalised in Natural Language Processing (NLP) and what this reveals about the possibilities and limits of considering a plurality of cultural backgrounds in technological design. It proposes that cultural alignment cannot be achieved only by adding more examples of "other cultures", rather it requires plural epistemologies: allowing multiple, locally grounded ways of knowing. To analyze how this plurality of knowing can be addressed in NLP, the pape

Confidence calibration in large language models (LLMs) is commonly evaluated by comparing predicted confidence with observed accuracy. However, such approaches do not model item difficulty, making it difficult to interpret discrepancies and to determine whether model confidence reflects genuine self-assessment or is merely a byproduct of the response generation process. To address this, we adopt a Rasch model-based latent ability framework and a metacognitive perspective, and propose Latent Conf

Foundation models have achieved remarkable performance across medical question answering, imaging, and electronic health record (EHR) tasks, yet reliable clinical deployment remains challenging due to limited interpretability, vulnerability to distribution shift, and weak alignment with clinician reasoning. We argue that these limitations arise because existing approaches prioritize representation learning while treating patient comparison as an emergent property rather than a primary source of

Streaming text-to-speech synthesis in cascaded LLM-TTS systems still faces latency challenges as most TTS models require full context before initiating generation. We present S5-TTS, a streaming variant of T5-TTS that enables low-latency, word-by-word incremental speech synthesis through encoder-decoder language modeling and monotonic alignment learning. S5-TTS begins generating speech immediately after receiving the first few words, substantially reducing end-to-end response latency. To maintai

AI agents acting on behalf of users are constantly making decisions, and for users to trust their agents, those decisions must align with what they actually want. Privacy is an important alignment problem for agents: every message, post, or tool call an agent makes is a contextual judgment about what is appropriate to share, with whom, and under which conditions. Because such judgments depend on social expectations and norms, human judgment does not merely label privacy violations but also helps

Imitation learning has emerged as a powerful paradigm for learning visuomotor policies, but its generalisation and stability are limited by the scale and quality of demonstration data needed. A promising direction is to leverage more abundant but heterogeneous data sources, which differ in action space and often lack action labels altogether. Existing co-training approaches that combine heterogeneous data sources rely on heuristic and hand-engineered alignment techniques. In contrast, we argue t

Scaling wildlife monitoring for real-world conservation deployments requires automated analysis of smart sensors that operate under severe annotation scarcity. We propose leveraging expert knowledge of species activity patterns as an annotation-free validation signal for multimodal monitoring pipelines. We operationalize agreement as the alignment of independently derived hourly activity curves both with each other and with published behavioral priors-a three-way convergence that rules out share

Alignment from human feedback uses human judgments about model outputs to steer the behavior of language models after pretraining. When those judgments reflect conflicting views of desirable behavior, the learned objective becomes an aggregate determination of what the model should prefer. We survey recent work that has studied this aggregation problem through the lens of social choice theory. We illustrate how the social choice perspective helps identify failure modes in the feedback aggregatio

Medical image re-identification (MedReID) enables longitudinal patient linkage but remains vulnerable to shortcut learning and often produces decisions that clinicians cannot audit against named anatomy. We propose Graph-of-Differences (GoD), which grounds identity comparisons in explicit anatomical structure. Each image is represented as an anatomy graph whose nodes correspond to named anatomical regions; given an image pair, soft node correspondence is established, and differences are computed

Large Audio Language Models (LALMs) have demonstrated strong performance across a wide range of audio tasks. As they are increasingly deployed in real-world applications, ensuring their safety alignment has become more important. Although refusal mechanisms serve as a key safeguard by preventing LALMs from responding to harmful requests, they can also lead to {\em over-refusal}, where models incorrectly reject benign queries. This issue is especially challenging in the audio domain because speec

Dense sentence embeddings are fundamental to modern Retrieval-Augmented Generation (RAG) systems but suffer from a lack of interpretability due to feature superposition. This opacity hinders the alignment of retrieval processes with human intent, as the entangled representations are difficult to analyze or control. In this work, we propose a method to disentangle the dense representations of sentence transformers (e.g., E5) into human-interpretable concepts using Top-k Sparse Autoencoders (SAEs)

Emergent misalignment (EM) is a phenomenon in which models generalize with narrow fine-tuning, leading to broad (yet uneven) misalignment across evaluation questions. We study EM and its variability directly through the components of fine-tuning: training dynamics, model priors, and data. (1) We first explored how in-domain training loss relates to out-of-domain alignment scores across datasets and model families. Then, we tried to induce potential alternative local minima through different lear

Prior work has shown that in-context demonstrations can jailbreak language models, but it remains unclear how models interpret different types of compliance demonstrations. We study this by mixing benign compliance demonstrations (non-harmful request, helpful response) with harmful compliance demonstrations (harmful request, helpful response) and testing three hypotheses about how demonstration composition drives harmful compliance. Across four models, we find that benign and harmful demonstrati