Vision Language Models (VLMs) demonstrate strong qualitative visual understanding, but struggle with metrically precise spatial reasoning required for embodied applications. The agentic paradigm promises that VLMs can use a wide variety of tools that could augment these capabilities, such as depth estimators, segmentation models, and pose estimators. Yet it remains an open challenge how to realize this vision without solely relying on handcrafted prompting strategies or enforcing fixed, predefined tool pipelines that limit VLMs' ability to discover optimal tool-use patterns. Reinforcement Learning could overcome this gap, but has so far been limited to reasoning with a single visual tool due to the large search space in multi-tool reasoning. We introduce Double Interactive Reinforcement Learning (DIRL), a two-phase training framework where VLMs learn to coordinate multiple tools through interactive exploration and feedback. In the teaching phase, we combine demonstrations from a single tool specialist trained via interactive RL with traces from a frontier model using all tools. In the exploration phase, the model further refines multi-tool coordination through continued RL. Our model, SpaceTools, with tool-augmented spatial reasoning ability, achieves state-of-the-art performance on spatial understanding benchmarks (RoboSpatial-Home, BLINK, BOP-ASK) and demonstrates reliable real-world manipulation using a 7-DOF robot as a tool. DIRL provides substantial improvements over the vanilla SFT (+12% on RoboSpatial) and RL (+16% on RoboSpatial) baselines. Project page: https://spacetools.github.io/.

The discovery of Bell that there exist quantum correlations that cannot be reproduced classically is one of the most important in the foundations of quantum mechanics, as well as having practical implications. Bell's result was originally proven in a simple bipartite causal structure, but analogous results have also been shown in further causal structures. Here we study the only causal structure with six or fewer nodes in which the question of whether or not there exist quantum correlations that cannot be achieved classically was open. In this causal structure we show that such quantum correlations exist using a method that involves imposing additional restrictions on the correlations. This hence completes the picture of which causal structures of up to six nodes support non-classical quantum correlations. We also provide further illustrations of our method using other causal structures.

In democracies, major policy decisions typically require some form of majority or consensus, so elites must secure mass support to govern. Historically, elites could shape support only through limited instruments like schooling and mass media; advances in AI-driven persuasion sharply reduce the cost and increase the precision of shaping public opinion, making the distribution of preferences itself an object of deliberate design. We develop a dynamic model in which elites choose how much to reshape the distribution of policy preferences, subject to persuasion costs and a majority rule constraint. With a single elite, any optimal intervention tends to push society toward more polarized opinion profiles - a ``polarization pull'' - and improvements in persuasion technology accelerate this drift. When two opposed elites alternate in power, the same technology also creates incentives to park society in ``semi-lock'' regions where opinions are more cohesive and harder for a rival to overturn, so advances in persuasion can either heighten or dampen polarization depending on the environment. Taken together, cheaper persuasion technologies recast polarization as a strategic instrument of governance rather than a purely emergent social byproduct, with important implications for democratic stability as AI capabilities advance.

Watermarking aims to embed hidden signals in generated text that can be reliably detected when given access to a secret key. Open-weight language models pose acute challenges for such watermarking schemes because the inference-time interventions that dominate contemporary approaches cannot be enforced once model weights are public. Existing watermaking techniques for open-weight models, such as the recently proposed GaussMark, typically rely on small modifications to model weights, which can yield signals detectable to those equipped with a secret key, but achieving detection power comparable to inference-time watermarks generally requires weight perturbations that noticeably reduce generation quality. We introduce MarkTune, a theoretically principled, on-policy fine-tuning framework that treats the GaussMark signal as a reward while simultaneously regularizing against degradation in text quality. We derive MarkTune as an improvement on GaussMark and demonstrate that MarkTune consistently improves the quality-detectability trade-off over GaussMark by steering finer-grained, watermark-aware weight updates within the model's representation space while preserving generation quality. Empirically, we show that MarkTune pushes the quality-detectability frontier of GaussMark close to that of inference-time watermarking, remains robust to paraphrasing and fine-tuning attacks, and exhibits strong generalization: a model fine-tuned on one dataset retains substantial watermark detection power on unseen datasets. Together, these results establish MarkTune as a general strategy for embedding robust, high-quality watermarks into open-weight LMs.

Training with differential privacy (DP) provides a guarantee to members in a dataset that they cannot be identified by users of the released model. However, those data providers, and, in general, the public, lack methods to efficiently verify that models trained on their data satisfy DP guarantees. The amount of compute needed to verify DP guarantees for current algorithms scales with the amount of compute required to train the model. In this paper we design the first DP algorithm with near optimal privacy-utility trade-offs but whose DP guarantees can be verified cheaper than training. We focus on DP stochastic convex optimization (DP-SCO), where optimal privacy-utility trade-offs are known. Here we show we can obtain tight privacy-utility trade-offs by privately minimizing a series of regularized objectives and only using the standard DP composition bound. Crucially, this method can be verified with much less compute than training. This leads to the first known DP-SCO algorithm with near optimal privacy-utility whose DP verification scales better than training cost, significantly reducing verification costs on large datasets.

Traffic state estimation (TSE) becomes challenging when probe-vehicle penetration is low and observations are spatially sparse. Pure data-driven methods lack physical explanations and have poor generalization when observed data is sparse. In contrast, physical models have difficulty integrating uncertainties and capturing the real complexity of traffic. To bridge this gap, recent studies have explored combining them by embedding physical structure into Gaussian process. These approaches typically introduce the governing equations as soft constraints through pseudo-observations, enabling the integration of model structure within a variational framework. However, these methods rely heavily on penalty tuning and lack principled uncertainty calibration, which makes them sensitive to model mis-specification. In this work, we address these limitations by presenting a novel Physics-Embedded Gaussian Process (PEGP), designed to integrate domain knowledge with data-driven methods in traffic state estimation. Specifically, we design two multi-output kernels informed by classic traffic flow models, constructed via the explicit application of the linearized differential operator. Experiments on HighD, NGSIM show consistent improvements over non-physics baselines. PEGP-ARZ proves more reliable under sparse observation, while PEGP-LWR achieves lower errors with denser observation. Ablation study further reveals that PEGP-ARZ residuals align closely with physics and yield calibrated, interpretable uncertainty, whereas PEGP-LWR residuals are more orthogonal and produce nearly constant variance fields. This PEGP framework combines physical priors, uncertainty quantification, which can provide reliable support for TSE.

Animals with foveated vision, including humans, experience microsaccades, small, rapid eye movements that they are not aware of. Inspired by this phenomenon, we develop a method for "Artificial Microsaccade Compensation". It can stabilize video captured by a tailless ornithopter that has resisted attempts to use camera-based sensing because it shakes at 12-20 Hz. Our approach minimizes changes in image intensity by optimizing over 3D rotation represented in SO(3). This results in a stabilized video, computed in real time, suitable for human viewing, and free from distortion. When adapted to hold a fixed viewing orientation, up to occasional saccades, it can dramatically reduce inter-frame motion while also benefiting from an efficient recursive update. When compared to Adobe Premier Pro's warp stabilizer, which is widely regarded as the best commercial video stabilization software available, our method achieves higher quality results while also running in real time.

Vision-Language Models (VLMs) deployed in safety-critical applications such as autonomous driving must handle continuous visual streams under imperfect conditions. However, existing benchmarks focus on static, high-quality images and ignore temporal degradation and error propagation, which are critical failure modes where transient visual corruption induces hallucinations that persist across subsequent frames. We introduce DIQ-H, the first benchmark for evaluating VLM robustness under dynamic visual degradation in temporal sequences. DIQ-H applies physics-based corruptions including motion blur, sensor noise, and compression artifacts, and measures hallucination persistence, error recovery, and temporal consistency through multi-turn question-answering tasks. To enable scalable annotation, we propose Uncertainty-Guided Iterative Refinement (UIR), which generates reliable pseudo-ground-truth using lightweight VLMs with uncertainty filtering, achieving a 15.3 percent accuracy improvement. Experiments on 16 state-of-the-art VLMs reveal substantial robustness gaps: even advanced models such as GPT-4o achieve only a 78.5 percent recovery rate, while open-source models struggle with temporal consistency at less than 60 percent. DIQ-H provides a comprehensive platform for evaluating VLM reliability in real-world deployments.

Tokenizer adaptation plays an important role in transferring pre-trained language models to new domains or languages. In this work, we address two complementary aspects of this process: vocabulary extension and pruning. The common approach to extension trains a new tokenizer on domain-specific text and appends the tokens that do not overlap with the existing vocabulary, which often results in many tokens that are unreachable or never used. We propose continued BPE training, which adapts a pre-trained tokenizer by continuing the BPE merge learning process on new data. Experiments across multiple languages and model families show that this approach improves tokenization efficiency and leads to better utilization of added vocabulary. We also introduce leaf-based vocabulary pruning, which removes redundant tokens while preserving model quality. Together, these methods provide practical tools for controlled vocabulary modification, which we release as an open-source package.

Online platforms connect users with relevant products and services using ads. A key challenge is that a user's search query often leaves their true intent ambiguous. Typically, platforms passively predict relevance based on available signals and in some cases offer query refinements. The shift from traditional search to conversational AI provides a new approach. When a user's query is ambiguous, a Large Language Model (LLM) can proactively offer several clarifying follow-up prompts. In this paper we consider the following: what if some of these follow-up prompts can be ``sponsored,'' i.e., selected for their advertising potential. How should these ``suggestion slots'' be allocated? And, how does this new mechanism interact with the traditional ad auction that might follow? This paper introduces a formal model for designing and analyzing these interactive platforms. We use this model to investigate a critical engineering choice: whether it is better to build an end-to-end pipeline that jointly optimizes the user interaction and the final ad auction, or to decouple them into separate mechanisms for the suggestion slots and another for the subsequent ad slot. We show that the VCG mechanism can be adopted to jointly optimize the sponsored suggestion and the ads that follow; while this mechanism is more complex, it achieves outcomes that are efficient and truthful. On the other hand, we prove that the simple-to-implement modular approach suffers from strategic inefficiency: its Price of Anarchy is unbounded.