ResearchNvidia research: harness design, not just model choice, drives success on long‑horizon tasksNvidia researchers report that a tailored agent harness — including a supervisory “boss” component and memory-aware handling — was decisive in getting Claude Opus 5 to a perfect score on the interactive reasoning benchmark ARC‑AGI‑3.3 min read
ResearchDeepMind partners with game developers to build generalist gaming agents and research in the EVE universeDeepMind is collaborating with multiple game studios to develop generalist agents that can see, understand and act in existing games without code changes, advancing both gameplay and AI research.5 min read
ResearchEmbedding Mobility into POIs: Enriching Place Representations for Language ModelsResearchers propose Mobility-Embedded POIs (ME-POIs), a framework that combines anonymized, aggregated mobility patterns with text-based embeddings to create richer vector representations of places.4 min read
ResearchOuter Biosciences trains AI on living human skin to speed cosmetic ingredient discoveryOuter Biosciences, founded in 2022, keeps surgically discarded human skin alive for weeks and combines those experiments with machine learning to discover cosmetic ingredients.5 min read
ResearchBenchmark optimization inflates ASR scores by teaching models to match test-specific transcriptsOpen-source speech recognition models can learn to reproduce quirks of public benchmarks instead of faithfully transcribing audio, inflating reported performance.4 min read
ResearchPew Research: Over a Third of Post‑ChatGPT Web Pages Show Signs of AI AuthorshipA Pew Research Center study finds that more than 35% of English‑language web pages published after the release of ChatGPT display signs of having been written or substantially edited by AI.3 min read
ResearchMerck and Moderna's Personalized mRNA Vaccine Halves Melanoma Recurrence in 1,000-Patient TrialMerck and Moderna reported that their personalized mRNA cancer vaccine, intismeran autogene, met primary endpoints in a 1,000-patient melanoma trial, halving recurrence and reducing spread by 59% without new safety signals.2 min read
ResearchMIT study finds generative AI images often can't be traced to specific training imagesA study by MIT’s Computer Science and Artificial Intelligence Laboratory identifies “attribution decay,” the tendency for images produced by generative models trained on very large datasets to lose traceability to individual training images.2 min read
ResearchEfficient federated multimodal training with NVIDIA FLARE and FedUMMNVIDIA FLARE and the FedUMM framework demonstrate approaches to federated training of vision-language and unified multimodal models by minimizing what model state crosses the network and by streaming/externally storing large updates.5 min read
ResearchClaude artificial intelligence designed de novo protein binders for 14 of the 15 targetsThe artificial intelligence called Claude designed de novo protein-binding molecules, and based on a human expert prompt autonomously and successfully designed binders for 14 of the 15 targets tested.1 min read
ResearchMost U.S. Young Adults Say They Are More Concerned Than Excited About AIA Pew Research Center survey released Tuesday finds that 55% of U.S.2 min read
ResearchIndependent dataset shows big AI firms' usage reports omit many sensitive usesA team led by Stanford and MIT researchers built the AI Observatory, an independent platform aggregating 24,521 real user–AI conversations from 2023–2025 to analyze how people use models like ChatGPT, Claude, Gemini and Grok.4 min read
