Alignment, interpretability, red-teaming, bias and privacy: the research on what these systems do when they misbehave.
arXiv:2609.14421v1 Announce Type: cross Abstract: Recent advances in multimodal foundation models have intensified the need to understand how different modalities share, preserve, and complement info...
arXiv:2610.14624v1 Announce Type: cross Abstract: Interpreting Earth system conditions is central to understanding environmental risks and informing decisions across weather- and climate-sensitive se...
arXiv:2610.14615v1 Announce Type: cross Abstract: We study autoregressive token generation through the notion of sequential samplers: A circuit that samples the next symbol only given the previous on...
arXiv:2610.14705v1 Announce Type: cross Abstract: Large language models have advanced medical reasoning, but clinically grounded reasoning in traditional Chinese medicine (TCM) remains challenging be...
arXiv:2610.14723v1 Announce Type: cross Abstract: A method integrating policy gradient with fuzzy inference has been proposed. This method, hereafter referred to as the base model, has high flexibili...
arXiv:2610.15127v1 Announce Type: cross Abstract: Traffic sign recognition plays a crucial role in the development of autonomous cars to reduce the accident rate and promote road safety. It has been...
arXiv:2610.15359v1 Announce Type: cross Abstract: Deep classifiers deployed in safety-critical services like autonomous-driving perception remain vulnerable to small adversarial perturbations. Certif...
arXiv:2610.15961v1 Announce Type: cross Abstract: Emergent misalignment (EM) occurs when fine-tuning a language model to a narrow task makes it misaligned on unrelated ones. Because organizations now...
arXiv:2610.16213v1 Announce Type: cross Abstract: Embedded platforms increasingly host safety-related streaming functions next to less critical ones. Whether a safety architecture pattern, such as a...
arXiv:2610.16622v1 Announce Type: cross Abstract: With the advancement in language modeling, research on adversarial images has recently shifted from image classifiers to vision-language models (VLMs...
arXiv:2610.16790v1 Announce Type: cross Abstract: Language model APIs can expose token-specific biases while withholding logits and token probabilities. Existing bias-based extraction recovers relati...
arXiv:2610.17685v1 Announce Type: cross Abstract: Enterprise agents face safety risks that cannot be determined from individual actions alone. Existing safety benchmarks largely evaluate action-local...
arXiv:2610.18066v1 Announce Type: cross Abstract: Reliable uncertainty evaluation is essential for assessing the trustworthiness of machine learning (ML) predictions, particularly in safety-critical...
arXiv:2610.18243v1 Announce Type: cross Abstract: Given the growing global burden of liver tumors and diffuse liver diseases, non-contrast computed tomography (NCCT) is increasingly used for opportun...
arXiv:2610.18294v1 Announce Type: cross Abstract: Releasing a neural network checkpoint can leak information about its training data, and existing reconstruction results rely on gradient-descent trai...
arXiv:2610.18415v1 Announce Type: cross Abstract: Large Language Models (LLMs) are aligned to human preferences, after pre-training and supervised fine-tuning steps. However, standard RL alignment st...
arXiv:2610.18609v1 Announce Type: cross Abstract: Direct Preference Optimization (DPO) has become a widely adopted approach for aligning vision-language models (VLMs), yet the security of multimodal...
arXiv:2610.18612v1 Announce Type: cross Abstract: System One models such as Jev change how software makes decisions by mapping open-ended semantic contexts into structured judgments that directly dri...
arXiv:2610.18694v1 Announce Type: cross Abstract: Vision-language models (VLMs) struggle with metric and multi-view relationships that remain implicit in RGB observations. Geometry foundation models...
arXiv:2511.17937v2 Announce Type: replace Abstract: Alignment faking is a form of strategic deception in AI in which models selectively comply with training objectives when they infer that they are i...