News

The latest published briefings, with original sources and the limits of each claim.

Browse all published stories or choose a period. Dates below are the original sources’ publication dates. New stories appear after their source-support checks.

Week starting 9 Nov 2026

9 Nov 2026 – 15 Nov 2026
ResearchRSS briefIntroductoryOpenAI launches science blogOpenAI says it has started a science blog that will share research findings, practical methods, and firsthand accounts from researchers using AI in scientific discovery.OpenAI · 15 Nov 2026Only the publisher RSS excerpt was available; the full article was not reviewed.ResearchIntroductoryNVIDIA walkthrough shows agent-built robot pick-and-place in Isaac Sim using SimReady assetsA new tutorial combines a UR17G arm, Robotiq 2F-140 gripper, and beverage can from SimReady Central, with an AI agent generating the task code.huggingface.co · 13 Nov 2026The article is a single-source walkthrough by NVIDIA; no independent party has reproduced the pick-and-place results.ResearchIntroductoryAllen AI's XYBench finds chatbots answer the question asked, not the problem behind itAI2 and collaborators released XYBench, an 8,115-question benchmark testing whether AI assistants spot the misconception hidden in a user's request and redirect them to what they actually need. Five leading models trailed human answers.huggingface.co · 13 Nov 2026All findings come from the benchmark's authors; no independent evaluation is cited.Models & toolsRSS briefIntroductoryOpenAI introduces ChatGPT Wallet for agent purchasesOpenAI announced ChatGPT Wallet, described as a secure way for agents in ChatGPT Work, Codex, and dots to complete purchases with user approval.OpenAI · 13 Nov 2026Only the publisher RSS excerpt was available; the full article was not reviewed, so availability, pricing, and rollout details are unconfirmed.ResearchPractitionerllama.cpp reaches version 1.0.0 with stable API promisesThe ggml-org team behind llama.cpp has released version 1.0.0 of its libraries, committing to clearer semantic versioning so downstream developers can see when API or ABI changes break compatibility.huggingface.co · 13 Nov 2026The announcement comes from the ggml-org team itself; no independent testing or third-party verification of the stability claims or backend coverage is included.ResearchRSS briefIntroductoryShield uses OpenAI agents to hand off legacy portal workOpenAI's customer story says Shield relies on its technology so IT technicians can pass along work in legacy portals, with Forge agents navigating systems.OpenAI · 13 Nov 2026Only the publisher RSS excerpt was available; the full article was not reviewed, so the claim reflects OpenAI's own description and is not independently confirmed.ResearchIntroductoryWinfree Oscillatory Neural Network scales synchronization-based AI to ImageNetA new arXiv paper proposes WONN, an oscillatory neural network whose authors say is the first synchronization-based architecture to compete on ImageNet-1K while using far fewer parameters on reasoning tasks.arXiv · 13 Nov 2026The evidence is limited to the paper's abstract; full results, experimental setup and comparisons are not available for verification.ResearchIntroductorySkill Adapter paper proposes personalized skill routing for LLM agentsA new arXiv paper describes a system that adapts shared agent skills to individual users and routes them during execution, reporting gains over SkillRL on three benchmarks.arXiv · 13 Nov 2026The evidence is the paper's abstract only; evaluation details, baselines beyond SkillRL, and the basis for the reported figures cannot be checked from the excerpt.ResearchIntroductoryLLM agents generate annotated trajectory-anomaly dataset with map-constrained routingAn arXiv paper proposes an end-to-end framework that synthesizes human trajectory anomalies at scale, using LLM agents, map-constrained routing and GPS noise simulation to create labeled ground-truth data.arXiv · 13 Nov 2026Evidence is limited to the paper's abstract; the full methodology, experiments and results are not covered.ResearchIntroductoryStudy finds random matrix theory metrics for neural network layers can be arbitraryAn arXiv paper argues that data-free metrics based on random matrix theory do not reliably reflect model quality, because reparameterised networks can keep performance while metrics change.arXiv · 13 Nov 2026The evidence is the paper's abstract only; the full article, including methods, experiments and peer review status, is not covered, and no independent confirmation of the findings is available.Models & toolsIntroductoryStudy links GPT assistance to faster GitHub pull request resolutionResearchers report that pull requests assisted by GPT were resolved in a median of 9 hours versus 23 hours, based on 25,473 GitHub pull requests across 9,254 projects.arXiv · 13 Nov 2026The evidence is the paper's abstract only, so the full methodology, dataset construction, and robustness checks cannot be verified from this excerpt.Agents & MCPIntroductoryResearchers define four conditions for LLM agents that survive process death without a hostAn arXiv paper proposes that plugin-based agents can run across peer processes with no host, provided four conditions on state, records, recovery and providers hold.arXiv · 13 Nov 2026Evidence is limited to the paper's abstract; task design, fault-injection method and full experimental detail are not available for verification.ResearchIntroductoryResearchers propose society-centered framework for judging AI governance by recourse, not principlesA new arXiv paper introduces a framework assessing whether people can get notice of AI-influenced decisions, challenge them, and see corrections, tested against North Carolina and 49 state documents.arXiv · 13 Nov 2026The source is an abstract only; the full paper's methods, data, and peer-review status are not visible, and the findings are not independently confirmed.ResearchIntroductoryCompact neural network targets P300 brain-computer interface decoding with 5,100 parametersResearchers describe a small deep learning model that combines xDAWN spatial filtering, a fixed filter bank and convolutional layers to detect P300 EEG responses, reporting 5,100 trainable parameters and benchmark results on two public datasets.arXiv · 13 Nov 2026The evidence is the paper's abstract only; full methods, datasets handling and peer review status are not visible in the supplied excerpt.ResearchIntroductoryDashBench finds Video LLMs fall short on rare driving corner casesA new arXiv paper introduces DashBench, a benchmark of 1,709 curated driving videos used to test 25 Video LLMs on rare, safety-critical road events, reporting that every model scored below the level needed for such applications.arXiv · 13 Nov 2026The evidence is the paper's abstract only; the full methodology, per-model results, and how the safety threshold was defined cannot be verified from this excerpt.Agents & MCPIntroductoryCortex-FL uses AI agents to organize federated learning collaborationsA new arXiv paper describes an agent-native federated learning system in which AI agents translate natural-language goals into shared training tasks, find collaborators, and negotiate participation agreements.arXiv · 13 Nov 2026Evidence is limited to the paper's abstract; the full article was not reviewed, so technical details and evaluation results are not covered here.ResearchIntroductoryResearchers propose prototype-based framework linking vision models to brain activityAn arXiv abstract describes a unified method that replaces black-box visual features with interpretable prototypes, aiming to connect vision foundation models to fMRI responses across several analysis tasks.arXiv · 13 Nov 2026Coverage is limited to the paper's abstract; the full article's methods, data and results were not available for review.ResearchIntroductoryLGPO drops importance ratios from LLM reinforcement learning, cutting training timeResearchers propose Likelihood-Gated Policy Optimization, a reinforcement learning method for post-training large language models that removes old-policy likelihood computation while keeping training stable, according to their arXiv abstract.arXiv · 13 Nov 2026The evidence is the paper's abstract only; full experimental details, model sizes, benchmarks and peer review status are not available in the supplied passage.ResearchIntroductoryGraphSkillAA ties agent skill-graph updates to attribution, validation and rollbackThe arXiv paper GraphSkillAA proposes updating an agent's skill graph after failures, using error attribution, targeted retesting and rollback while model parameters stay frozen.arXiv · 13 Nov 2026The evidence is the paper's abstract only; experimental setup, baselines and comparison details are not available in the supplied passages, and the reported results are the authors' own, not independently confirmed.ResearchIntroductoryLoRA fine-tuning study finds reasoning traces add little for Qwen3-8B finance tasksAn arXiv paper compares three ways of adapting Qwen3-8B to financial data with LoRA and reports that simple response-only training performs best overall.arXiv · 13 Nov 2026The source is the paper's abstract only, so experimental details, datasets, and full results are not available for verification.

Subscribe with RSS · How the checks work