Natural Language Processing news and analysis on Web Pulse: 5440 curated articles tracking the latest Natural Language Processing developments, tools, and research, updated continuously from vetted sources.
Researchers formalized a vulnerability they call Verdict-Preserving-Unfaithfulness (VPU), in which mathematical solvers in neurosymbolic systems verify reasoning correctness but cannot detect whether a formal translation…
Cohere released North Small Translate, an open-weights mixture-of-experts machine translation model that scores 83.6 on the WMT26 All Languages benchmark, which Cohere says outperforms DeepL NextGen (81.37) and Google Tr…
OpenAI launched GPT-Live-1 in its API on September 10, giving developers the full-duplex voice model that has powered ChatGPT Voice since 2025 at $0.05 per minute for the voice layer. OpenAI reported the model improves f…
The University at Albany (SUNY) developed the Research Highlighter-MatchMaker Project, an NLP-driven software platform that matches researchers with funding opportunities and collaborators through a searchable portal and…
Google said its compliance with the European Union's Digital Markets Act produced the largest reduction in search quality in Europe, according to Search Engine Roundtable's weekly recap. Google also confirmed a temporary…
AETHERCORE added a Five Paths view to its GenesisCore astrology product on September 9, 2026, using 386 fixed rules and a prepared explanation dictionary rather than a new LLM call for the judgment layer. The feature sep…
A developer built NexusOS, a locally running AI-powered operating environment that responds to the wake word "Hey Nexus" and executes voice or text commands to control a computer. The system combines Whisper speech recog…
A Hacker News user asked how realistic a modern-day natural language understanding (NLU) system would be for tool-calling tasks, proposing a targeted 1B to 3B parameter model that handles tool selection without code gene…
Nano Singapore launched Jenny, an AI-powered wellness advisor that uses real-time voice conversation and a lifelike digital avatar to move shoppers from a vague health concern to a specific product recommendation without…
A developer argues that most enterprise RAG pipeline failures originate at the document parsing stage of ingestion, before any query runs, rather than in the LLM or retrieval layers teams typically tune. The writeup docu…
Prompt engineering — designing and optimizing the text inputs given to large language models — can let a less advanced model such as GPT-3.5 outperform a more sophisticated model like GPT-4 when the latter receives a poo…
GoSumo, an open-source project, has launched an AI-powered smart-routing engine that consolidates customer messages from WhatsApp Business, Instagram Direct, SMS, web chat, and email into a single unified inbox for small…
Cohere released North Small Translate, an open-weight sparse Mixture-of-Experts translation model with 218B total and 25B active parameters covering 50 languages, which scores 83.6 on Cohere's WMT26 all-languages evaluat…
GoSumo has launched a smart routing feature that consolidates inbound customer conversations from WhatsApp, Instagram, SMS, web chat, and email into a single AI-driven interface for India SMBs. The platform forwards mess…
Researchers introduced X-AuT, a progressive framework that selects layer combinations to compress audio-encoder depth for speech large language models, addressing the deletion and premature end-of-sequence errors caused …
Late-interaction retrieval remains the state-of-the-art for visual document search but its accuracy comes at a storage cost, since existing compression methods retain only a subset or local average of the roughly 1,000 v…
Swiftime launched an AI-powered time tracking app that converts a single typed or spoken sentence into a structured time entry with client, project, duration, and date filled in automatically. The app offers speech input…
Researchers introduced LLM-Anchored Paralinguistic Enrichment (LAPE), a method that combines LLM-derived linguistic representations with paralinguistic speech cues to detect Alzheimer's disease, achieving state-of-the-ar…
A study of 93 Urdu stories generated by three contemporary LLMs — GPT-5.1, Qwen-3-Max, and DeepSeek-3.1 — found that the models frequently make basic grammar and semantic errors, produce incoherent and repetitively unnat…
A training-free framework combining reasoning and retrieval over Wikipedia improves multilingual multimodal entity linking by 6.9% overall and up to 23.3% on rare-entity slices on the MERLIN benchmark, according to an ar…