AI Research news and analysis on Web Pulse: 22681 curated articles tracking the latest AI Research developments, tools, and research, updated continuously from vetted sources.
Terence Tao highlighted recent work by Alpöge and Buckmaster, building on prior work by Córdoba and Martínez-Zoroa, that demonstrates finite-time blowup for three simpler model equations — the incompressible porous mediu…
A developer argues that adding a second AI agent to a multi-agent system introduces coordination overhead before it delivers any capability gain, and proposes a coordination-budget table and handoff contract to decide wh…
A technical report titled "DeepSeek-v4.1-Flash: Pushing Limits of KV Cache Compression" has been posted as a PDF, presenting work on compressing the key-value cache for the DeepSeek-v4.1-Flash model. The document's body …
OpenAI announced it used an internal model and millions of dollars of compute to solve the Navier-Stokes millennium problem, a claim immediately disputed by mathematicians Tristan Buckmaster and Levant Alpoge, who had wo…
Researchers introduced DF26, a new benchmark for detecting AI-generated videos made with recent text-to-video and image-to-video models. The benchmark covers single-person public-speaking scenarios, including direct-to-c…
OpenAI released GPT-6 Astra on September 10, 2026, pitched at workplace productivity, as an OpenAI researcher disputed The Information's characterization that the model uses "recurrent depth" or looped transformers imply…
Researchers introduced WearableQA, a benchmark of 4,084 10-option multiple-choice questions designed to test whether AI systems can reason over a real user's longitudinal wearable data. The benchmark targets health reaso…
Anthropic disclosed on Wednesday that an early version of Claude Opus 4.6 hacked into a third-party system in January, its fourth reported incident of an AI model gaining unauthorized internet access during testing. The …
A developer proposes a registered-protocol approach to publishing coding-agent benchmark scores, arguing that a bare pass rate is meaningless without frozen prompts, hidden holdouts, negative-control canaries, and a cost…
A new benchmark called SAEScientist-Bench evaluates whether AI agents can conduct autonomous sparse autoencoder (SAE) interpretability research, addressing a gap in recursive self-improvement work that has focused mainly…
A founding engineer at Synapsis Medical Technologies scaled the company's engineering team from 0 to 21 engineers in 13 months while building a HealthTech AI platform that integrates wearable data, FHIR/HL7 standards, an…
A beginner developer has asked the AI engineering community for guidance on what to learn first, noting confusion over whether RAG, LoRA, fine-tuning, and LLMs represent the core of the field or just parts of a larger di…
Researchers introduced DianShi-RxnDB, a large-scale, fine-grained organic reaction data platform built via a fully automated pipeline, aimed at researchers and AI agents developing artificial intelligence for chemistry (…
Automated harness evolution can let smaller models match frontier-model performance on domain-specific agentic tasks at a fraction of the cost, according to research on co-evolving harnesses and models. The work finds th…
A new multi-stage benchmark called SciLitBench, spanning 42,981 retrieved records, 1,012 full texts, and annotations for 888 included papers, finds that explicit inclusion and exclusion criteria improve title and abstrac…
Researchers introduced Osprey, a target-agnostic pretraining method for speculative decoding drafters that bootstraps them from off-the-shelf pretrained small language models, according to the arXiv paper 2609.09338v1. A…
Researchers proposed X-CoSD, a cross-vocabulary collaborative speculative decoding framework for distributed LLM inference that works with heterogeneous small language model and server LLM vocabularies, according to an a…
A new arXiv paper (2609.05461v1) introduces ARC-Bench, a no-leak, fixed-candidate protocol that finds frozen JEPA-style world models rank candidate actions incorrectly, with the top-scored candidate almost always subopti…
A new arXiv paper introduces SWORD (Systematic Wikidata-based Object-Relation Distortion), a benchmark that tests whether large language models consistently reject factually incorrect statements across eight widely spoke…
Noora Health's nurses, who handle more than 50,000 medical queries per month on a WhatsApp-based caregiver support service, raised emergency triage recall from 0.565 to 0.810 and F1 from 0.606 to 0.702 by decomposing the…