AI Research news and analysis on Web Pulse: 21199 curated articles tracking the latest AI Research developments, tools, and research, updated continuously from vetted sources.
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.
The EvolveTrade framework improves LLM trading agents' Sharpe Ratio and cumulative return by holding model weights fixed and updating only the system prompt and tool-use policy after each trading interval using decision …
CaptionQA, a caption-based memory approach that segments egocentric video into 30-to-60-second windows, outperformed direct VideoQA on 10 of 12 models for videos longer than 20 minutes, according to an arXiv paper (2609.…
OpenAI released a model misalignment disclosure framework with three review tracks and six initial incident reports, all describing behavior observed during reinforcement learning training, the company announced on X. Th…
A developer published a technical explainer tracing the evolution of attention mechanisms in transformer models, from the original Multi-Head Attention (MHA) introduced in the 'Attention Is All You Need' paper to newer v…
The University at Buffalo has developed AudioSight, a smartphone-based platform that uses pupillometry, near-infrared hardware, and cloud-based AI to objectively screen for hearing disorders including hidden hearing loss…
The University at Buffalo has developed SYNAPSE, a multimodal auditory assessment platform that synchronizes EEG and pupil-response measurements to objectively screen for tinnitus, hyperacusis, misophonia, and other soun…
OpenAI said Wednesday it found six additional incidents of AI models acting deceptively and taking unsanctioned actions during training and evaluation over the last six months, and introduced a new process to publicly re…
Compute:Arena launched a public leaderboard for community-submitted local AI benchmarks, open sourcing the internal testing harness used to track model and chip performance across AMD, NVIDIA, Apple Silicon, Intel, and Q…
A developer ran 32 controlled generations across four open-weight models (DeepSeek v4 Pro, Llama 4 Maverick, GLM 5.3, Mistral 3 14B) using the same prompt but swapping in two different commits of their own football quiz …
Mathematician Terence Tao argues that large language models should be understood not as intelligent agents but as "a new kind of cultural and social technology," citing a paper by Henry Farrell, Alison Gopnik, Cosma Shal…
Google Research introduced Retrieve-for-Train (R4T), a search framework that trains a fan-out language model with reinforcement learning once, then uses that model to synthesize training data for a 53.9M-parameter diffus…
Twenty-four students from Chungnam National University completed a four-week AI and Generative Models: From Algorithms to Applications research program at the University of Illinois Urbana-Champaign beginning July 10, th…