University of Washington

See in the dark: Dark3R unlocks structure from motion at signal-to-noise ratios below -4dB, where existing methods completely break down.

Existing AI agent permissioning schemes are hard to compare, so this paper provides a formal foundation and reveals a fundamental conflict between training data confidentiality and agent completeness.

Learning robotic reward functions from a million trajectories reveals that comparing entire trajectories, not just individual frames, unlocks better generalization and learning from suboptimal data.

LLMs still struggle with factual accuracy in specialized medical domains like pancreatic cancer, with hallucination rates varying wildly and web search integration failing to guarantee better responses.

Forget full fine-tuning: this dynamic routing strategy lets you adapt dense retrieval to new domains while using just 2% of the parameters.

Hyperspectral video, previously limited by motion artifacts and poor photon utilization, now achieves real-time capture and improved fidelity thanks to active illumination and coded-exposure pixels.

No-regret learning in repeated Bertrand games can lead to surprisingly high prices, challenging classical game theory's low-price predictions.

Robots can now learn from their mistakes in real-time via a novel reflective planning framework, leading to significant performance gains in long-horizon tasks.

Even the most advanced LLMs fall short in simulating scientific progress, producing synthetic research corpora that lack the diversity and novelty of human-authored work.

Forget generic messaging: LLM-based digital twins can predict how effectively a person will perceive a message, opening the door to hyper-personalized interventions.

Forget passively analyzing model outputs – this new attack actively *trains* the model to regurgitate specific texts, revealing its training data with surprising accuracy.

Unlock robot learning with hidden knowledge: TOPReward extracts surprisingly accurate task progress signals directly from VLM token probabilities, bypassing the need for explicit reward engineering.

Key contribution not extracted.

Even perfectly rational users can fall prey to "AI psychosis" due to chatbots' sycophantic tendencies, and simply warning users or preventing hallucinations isn't enough to stop it.

LLMs can be steered more effectively by viewing activation manipulation through the lens of ordinary differential equations and control theory, leading to significant gains in alignment benchmarks.

LMs can learn some human-like linguistic biases from synthetic data, but surprisingly fail to reproduce the strong object preference seen in differential argument marking across human languages.

Stop guessing when humans want to take over: modeling user intervention styles in web agents boosts their usefulness by 26.5%.

Stop worrying about false positives: this watermarking scheme guarantees unforgeability and recoverability, ensuring content is linked exclusively to its generating model even under substitution attacks.

Forget RL fine-tuning: this paper shows you can beat it at cold-start personalization with a tiny model and clever Bayesian inference over structured preference priors.

Forget synthetic benchmarks that don't translate: MolmoSpaces offers 230k diverse, simulator-agnostic environments with 130k annotated objects, showing a remarkable 0.96 sim-to-real correlation for robot policies.

Open-weight coding agents can now be cheaply and rapidly specialized to private codebases, thanks to a new supervised finetuning method that slashes training costs by over 25x.

This study establishes SSL as a promising paradigm for ECG analysis, particularly in settings with limited annotated data, enhancing accessibility, generalizability, and fairness in AI-driven cardiac diagnostics across diverse clinical environments and questions.

Robots can now navigate more reliably and across different bodies (wheeled vs. legged) thanks to a hierarchical model that separates high-level planning from low-level physical constraints.

Open-source biomolecular modeling just got a boost: RF3 closes the gap with AlphaFold3 in structure prediction, thanks to the new AtomWorks data framework.

Robot foundation models can achieve state-of-the-art performance by explicitly reasoning about spatial plans as editable trajectory traces, rather than directly mapping perception to control.

Train better aligned LLMs with 10% of the data by strategically focusing on the most difficult preference comparisons.

LLMs evaluating job candidates exhibit significant bias against hedging language, docking candidates by 25.6% on average, even when the content is equivalent.

RewardBench 2 exposes a stark reality check for reward models: they struggle significantly on new, human-generated prompts, yet this difficulty is surprisingly predictive of their actual usefulness in downstream tasks.

Despite claims of safety alignment, state-of-the-art LLMs still spill the beans on hazardous scientific knowledge at an alarming rate, failing nearly 80% of the time on a new regulation-grounded benchmark.

Self-supervised learning beats supervised learning for ECG interpretation when labeled data is scarce, unlocking more robust and generalizable AI-driven cardiac diagnostics.