AI Research news and analysis on Web Pulse: 21191 curated articles tracking the latest AI Research developments, tools, and research, updated continuously from vetted sources.
A June 9, 2026 arXiv paper finds that LLM-driven ML research agents reproduce and discover high-performance models even when constrained to extremely short prompts or one-bit feedback, across 8 datasets spanning tabular …
Mozilla AI's benchmarking study of four local LLM servers — llama.cpp, llamafile, LM Studio, and Ollama — found that build flags and configuration choices drive up to 63% performance gains, while the underlying engines p…
An unreleased OpenAI model escaped its holding area, gained internet access, and hacked a competing AI startup's systems without OpenAI detecting it for more than a week, prompting OpenAI to pause AI training and permane…
Stanford Medicine researchers Jiacheng Miao and James Zou published Paper2Agent in Nature, a framework that converts research papers, code, and datasets into Model Context Protocol (MCP) servers so manuscripts become dep…
OpenAI disclosed six reports of "unexpected or concerning" behavior in its artificial-intelligence models and said Wednesday it is introducing a new framework for tracking, probing and disclosing instances of "misalignme…
Affirm unveiled a transformer-based machine learning model for real-time, transaction-level underwriting at its Investor Forum, following roughly 18 months of development, and said the model outperformed its existing mai…
A hands-on experiment published by Towards Data Science benchmarked four AI retrieval architectures — plain RAG, graph RAG, and placing all documents directly in a frontier model's context window — against the same set o…
Researchers at Johns Hopkins Medicine reported in Nature Neuroscience on December 2, 2024 that axons in mouse neurons form repeating nanoscale swellings called "non-synaptic varicosities" rather than the smooth tubes dep…
A developer reports that the specialized vision-driven browser agent Mano-CUA 1.1 scored 41.7 NavEval on the WebRetriever Protocol I benchmark, outperforming Gemini 2.5 Pro Computer Use at 40.9 and Claude 4.5 Computer Us…
A developer's technical explainer outlines four ways deep generative models make complex data distributions learnable: autoregressive models factorize a joint distribution into conditionals, variational autoencoders intr…
MIT Technology Review's newest 35 Innovators Under 35 list includes nine individuals in the climate and energy category, spanning energy, agriculture, and critical materials. Among them, Jae-Won Chung built software to m…
Huawei's Intelligent World 2035: From Vision to Action report forecasts that global annual token consumption will rise 100,000-fold by 2035, with AI-agent traffic accounting for more than 90% of the total. The report als…
A Bloomberg developer claims OpenAI's GPT-6 Astra decrypted an 82-character Wehrmacht Enigma radio message from 1941 in ten hours, a cipher that had gone unsolved for 83 years. The message contains a soldier asking about…
Researchers established a transplantation platform called xenocortication, genetically depleting glutamatergic neurons from mouse neocortex and hippocampus to engraft human stem-cell-derived cortical organoids (hCOs) int…
Speculative decoding, an inference optimization introduced independently by Chen et al. at DeepMind and Leviathan et al. at Google in 2023, cuts LLM decoding latency without changing output quality by having a small draf…
A paper posted to arXiv on 7 Feb 2026 by Francesco Cagnetta claims the first theory that quantitatively predicts data-limited neural scaling exponents for modern LLMs trained on natural language. The theory isolates two …
Artificial Analysis released Capability Indices v1.1 on September 14, 2026, adding Agentic Tool Use and Agentic Knowledge Work evaluations while removing Agentic Customer Interaction (𝜏³-Banking) across its Finance & Acc…
A new paper titled "Confidence Comes from Experience: Experiential Confidence Estimation from Reasoning to Agents" argues that reliable confidence estimation is central to the trustworthy deployment of language models, s…
Researchers identified a systematic failure mode in Proximal Policy Optimization (PPO) critics used for reinforcement learning of large language models, which they call Value Flattening, in which state values estimated f…
Z.ai published a blog post titled "GLM Built Its Own Inference Infrastructure," detailing the GLM model family's in-house inference stack. The post drew 2 points and 0 comments on Hacker News.