Alignment, interpretability, red-teaming, bias and privacy: the research on what these systems do when they misbehave.
arXiv:2603.02263v3 Announce Type: replace-cross Abstract: World models compress rich sensory streams into compact latent codes that anticipate future observations. We let separate agents acquire such...
arXiv:2603.10807v2 Announce Type: replace-cross Abstract: Existing LLM safety evaluations rely on binary attack-success rates and domain-agnostic taxonomies, leaving regulated Banking, Financial Serv...
arXiv:2603.11521v2 Announce Type: replace-cross Abstract: Unsupervised Camouflaged Object Detection (UCOD) remains a challenging task due to the high intrinsic similarity between target objects and t...
arXiv:2603.14602v3 Announce Type: replace-cross Abstract: Conversational assistants powered by large language models (LLMs) excel at tool-use tasks but struggle with adhering to complex, business-spe...
arXiv:2605.01609v2 Announce Type: replace-cross Abstract: We find that transformer concept representations systematically anti-concentrate in the spectral tail of the unembedding covariance, encoding...
arXiv:2605.22455v2 Announce Type: replace-cross Abstract: Real-world deployment of AI vision models is both fueled and limited by the data available for training and testing. Real datasets are sparse...
arXiv:2605.24614v2 Announce Type: replace-cross Abstract: Large language model (LLM) unlearning has emerged as a crucial post-hoc mechanism for privacy protection and AI safety, yet auditing whether...
arXiv:2605.27068v2 Announce Type: replace-cross Abstract: Social deduction games have become a popular testbed for probing reasoning, deception, coordination, and belief modeling in Large Language Mo...
arXiv:2605.28006v2 Announce Type: replace-cross Abstract: Understanding how LLMs reason is hindered by a practical asymmetry: while their generated outputs are observable, the underlying reasoning pa...
arXiv:2605.29089v2 Announce Type: replace-cross Abstract: Recent reinforcement learning (RL) post-training approaches primarily optimize the final output policy using sparse outcome-level rewards, wh...
arXiv:2605.29224v2 Announce Type: replace-cross Abstract: AI agents augment large language models with external tools such as web retrieval, enabling grounded and up-to-date responses. However, incor...
arXiv:2605.31556v2 Announce Type: replace-cross Abstract: Alignment teaches vision-language models (VLMs) to avoid expressing demographic biases, and when gender is clearly visible they largely succe...
arXiv:2607.02368v3 Announce Type: replace-cross Abstract: Language models prompted with cultural personas increasingly stand in for human respondents in cross-cultural research. Their responses separ...
arXiv:2607.11577v2 Announce Type: replace-cross Abstract: Conventional Graph Neural Networks (GNNs) couple feature transformation and neighborhood aggregation, which often renders them vulnerable to...
arXiv:2608.28934v1 Announce Type: new Abstract: Differential privacy (DP) has traditionally been used to provide theoretical upper bounds on an algorithm's stability to changing its training data. In...
arXiv:2608.29107v1 Announce Type: new Abstract: While modern generative models excel at modeling complex data, precise inference-time control in conditional generation remains a critical challenge. C...
arXiv:2608.29247v1 Announce Type: new Abstract: Deep neural networks remain highly vulnerable to adversarial perturbations, and adversarial training (AT) has become a widely used approach for improvi...
arXiv:2608.29647v1 Announce Type: new Abstract: To mitigate the time complexity of generative models, one-step generative models have recently emerged through direct mapping from noise to data in a s...
arXiv:2608.29674v1 Announce Type: new Abstract: Sharing tabular data in high-stakes domains is constrained by privacy regulations. Synthetic data offer a promising alternative, but deep generative mo...
arXiv:2608.29817v1 Announce Type: new Abstract: Neural Architecture Search (NAS) has so far rarely been applied to Mixture-of-Experts (MoE) models, and existing MoE designs leave the alignment betwee...