Alignment, interpretability, red-teaming, bias and privacy: the research on what these systems do when they misbehave.
Claims about AI safety reach audiences well beyond the AI community, yet many rely on opaque evidence or static assessments, when supporting evidence is accessible at all. We present the Systemic Risk...
OpenAI CEO Sam Altman addressed the United Nations Security Council, emphasizing the importance of AI safety, maintaining human control over artificial intelligence, and fostering international cooperation. He highlighted the need for global collaboration to manage AI risks and ensure responsible development. Altman’s remarks underscored the role of international governance in shaping the future of AI technology.
Reinforcement learning (RL) is widely used to sharpen reasoning in multimodal large language models (MLLMs), yet its effect on hallucination is uneven. We trace this to two weak points in the \emph{co...
Avatar-streaming systems are commonly evaluated with image and video quality assessment (IQA/VQA) metrics, implicitly treating visual fidelity as a proxy for communicative success. We test this assump...
PPG-to-vital-sign reconstruction turns a wrist-worn photoplethysmogram into clinical waveforms such as the ECG. Long-horizon multivariate time-series forecasting underpins planning in energy, weather,...
Speech technology penalizes some voices: recognition errs nearly twice as often for Black speakers, and accuracy declines for second-language accents and older speakers. We introduce TRIAD, an audit g...
Generative image models can produce highly realistic images, raising concerns about potential misuse in creating deceptive content. Current deepfake approaches face several challenges, including limit...
The paper introduces a new evaluation protocol for diffusion data‑point unlearning that tracks whether each target is forgotten, remains forgotten, or reappears during subsequent deletions. It identifies a failure mode called sequential reappearance, where an instance that was initially forgotten later returns to the memorized regime without re‑use of the deleted data or adversarial fine‑tuning. The study also finds that reappearing targets exhibit a sharper local denoising‑loss geometry after deletion than those that remain forgotten.
ForeDrive introduces a planning-relevant latent world model that is asymmetrically coupled to a Diffusion Transformer planner. The model learns multi‑horizon latent futures with a JEPA‑style world model, while planning gradients update the shared encoder and stop‑gradient routing trains the predictor with forecasting losses only. Gated visual fusion, future‑status injection, and Trajectory‑Adaptive Bias are used to guide trajectory generation without overriding current observations, achieving high performance on NAVSIM benchmarks using only front‑view images and pure imitation learning.
arXiv:2609.26733v1 Announce Type: new Abstract: The rapidly advancing capabilities of vision-language models (VLMs) present a systemic challenge to academic integrity. VLMs now allow students to bypa...
Semantic Self-Distillation (SSD) is a method that distills the semantic dispersion of sampled answers from large language models into lightweight student models. These students estimate a prompt-conditioned density before answer generation, providing a prompt-level uncertainty signal via entropy and an answer-level reliability measure through probability density. Experiments on TriviaQA and MMLU show that SSD matches the teacher’s uncertainty estimates while enabling additional tasks such as hallucination prediction, out-of-domain detection, and multiple-choice answer selection.
arXiv:2609.25576v1 Announce Type: cross Abstract: We study the estimation of a $K$-dimensional simplex from $N$ i.i.d.\ points sampled uniformly from its interior; the observations are convex combina...
The paper introduces ORDER, a fictitious-world benchmark designed to evaluate domain-adaptive embodied AI. ORDER consists of a synthetic 342,069-token corpus defining a self-consistent physics, a 500-question knowledge test (ORDER‑BENCH), and a compositional spatial task (ORDER‑SPATIAL) that requires ordering objects for safe manipulation. The benchmark demonstrates that models like GPT‑4.1 perform poorly without adaptation, while small models improve significantly after continual pre‑training, and that performance on ORDER‑SPATIAL better predicts real plan quality than knowledge-test accuracy.
The paper introduces a counterfactual observability framework for Netflix’s recommender systems, aiming to disentangle raw engagement signals—such as views and clicks—from confounding factors like content quality, model behavior, presentation bias, and audience reach. It proposes three stakeholder‑centered principles and measurement methods that reduce bias, assess relativity, and capture incrementality, applicable to both single‑stage and cascading recommender architectures. The framework is demonstrated through multiple production deployments, showing its effectiveness in enhancing observability across Netflix’s recommendation pipelines.
arXiv:2512. 21066v4 Announce Type: replace Abstract: Explainable artificial intelligence (XAI) reveals how explanatory variables relate to a response variable, yet communicating XAI outputs to laypersons remains difficult, limiting trust in AI-based predictions.
The study examines whether large language models (LLMs) can accurately simulate individual financial users by conducting a longitudinal paper‑trading experiment with 80 participants. Using a rolling next‑day prediction protocol, the researchers compared LLM predictions to a simple recent‑activity persistence baseline across multiple behavioral fidelity levels, from trade occurrence to asset selection and portfolio outcomes. Results show that no LLM consistently outperforms the baseline, with fidelity decreasing at finer behavioral granularity, and that recent trading history largely drives activity predictions while asset selection depends more on available evidence.
The paper introduces NavSecure, a vision‑based navigation framework that combines world models with safety‑focused decision making for autonomous vehicles operating over 5G networks. By predicting potential threats and generating safer routes, NavSecure reduces reliance on extensive real‑world trial‑and‑error learning and enables continuous autonomous adaptation. Experiments in simulation‑to‑real driving scenarios demonstrate superior performance in collision prevention and risk reduction compared to other end‑to‑end methods.
The study tests deliberately biased AI assistants and finds that such bias improves human performance on tasks like misinformation evaluation, financial investment, and graduate education compared to neutral AI. However, participants undervalue biased AI and overvalue neutral AI, even when performance is similar. When two AI biases flank a participant’s perspective, performance gains are maintained while reducing the perceived cost and one‑sided influence.
The paper introduces a black-box defense strategy for smart meter data that uses a proxy-guided hierarchical reinforcement learning framework to generate battery-based load-shaping policies. These policies inject realistic yet misleading appliance-level signatures into aggregate power signals, disrupting non-intrusive load monitoring attacks. Experiments on UK-DALE and REDD datasets show significant increases in appliance-level reconstruction error and reductions in attacker F1 scores across multiple unseen NILM models.
The paper investigates why diffusion models, unlike typical deep learning models, exhibit catastrophic overfitting when overparameterized. Through experiments on U‑Nets trained on CelebA and a random‑features theoretical analysis, it shows that the interpolation peak occurs at a model size proportional to the product of training samples and noise realizations, but the test loss starts to rise already at the number of samples, leading to memorization of the empirical score. Regularization techniques such as ridge penalties or early stopping can still make large models outperform smaller, unregularized ones.