Natural Language Processing

Stanford University researchers have developed an AI tool to analyze millions of hours of police body-camera footage, starting with a study of nearly 1,800 NYPD stops that revealed officers often failed to properly seek …

Dorset Council is trialing AI agents developed by Faculty to triage planning applications, summarize key information, and provide initial assessments for planning officers. The government aims to halve average processing…

Stanford researchers released the Stanford EDGAR Filings Dataset (SEFD), a 152 billion token reconstruction of SEC EDGAR filings from 1994 to present in a layout-faithful MultiMarkdown format, achieving over 99% structur…

Crosby, a legal-tech startup, released the Redline Bench to evaluate AI models on contract-review tasks such as redlining and risk-spotting. The benchmark addresses the challenge of defining correct outputs in legal edit…

Publishing more content is actively harming SEO performance as AI-driven search systems prioritize semantic clarity and entity authority over raw volume. Unlike traditional search engines that rewarded coverage, modern r…

An open-source Financial Intelligence Engine called BLUM is being built to combine financial data, news intelligence, sentiment analysis, and explainable AI reasoning into a transparent research platform. The project aim…

Seres Group unveiled a humanoid robot named Xiaosai for factory use, featuring visual recognition, autonomous greeting, and voice interaction. The company already operates factory variants Xiaosai 01 and Xiaosai 02 for c…

A developer launched an AI-powered Bible reader that presents multiple denominational perspectives on contested theological topics. The tool aims to help users explore different interpretations of scripture.

Mozilla's Firefox now includes on-device AI features such as local translation, PDF alt text generation, and sidebar chatbot integration, prioritizing privacy by avoiding cloud processing. The author praises the local AI…

Ollama 0.3.0 introduced a format parameter that enforces structured JSON output from local LLMs via constrained decoding, eliminating parsing failures caused by markdown wrappers or extraneous text. Tests with Gemma4 sho…

OpenAI expanded ChatGPT's voice input to support over 70 languages with automatic language detection on Android and iOS, and released three new developer models—GPT-Realtime-2, GPT-Realtime-Translate, and GPT-Realtime-Wh…

Researchers at Indiana University found that AI rewriting of radiology reports causes a 'slop paradox' where tasks that produce cleaner text for multimodal training actually degrade image-text alignment more than summari…

AI memory systems for LLM agents fail at scale due to structural precision problems in similarity search, not incidental issues. Testing across three embedding models showed mean retrieval precision of 0.09, with no impr…

A comparative study of classification models within the InferBERT framework for pharmacovigilance found that domain-specific pre-trained BioBERT outperformed larger models like Med-LLaMA in detecting causal adverse drug …

A study of a production enterprise LLM assistant with 110 agents and 584 tools found that routing accuracy degrades by 16-23 percentage points as the catalog scales, driven by retrieval and confusion gaps. Embedding-base…

Researchers at the University of Toronto introduced a novel method to evaluate second-order social bias in large language models (LLMs), where models exhibit bias in their judgments about biased content. Using a reasonin…

Researchers introduced NarrativeWorldBench, a benchmark for long-horizon co-creative audio drama, and N-VSSM, a latent world model that outperforms frontier LLMs like Claude Opus 4.5 on long-arc consistency by 71% in a w…

Researchers tested Word2Vec on Toki Pona, a constructed language with only ~130 words, using 1.4 million sentences. They found that non-core tokens like loanwords improved embedding quality by drawing similar words close…

Researchers have theoretically analyzed how deep transformers represent hierarchical structures in language, constructing models whose depth grows linearly with grammar depth and whose neuron count scales with derivation…

The MLLP-VRAIN research group submitted a cascaded system for the IWSLT 2026 Simultaneous Speech Translation task, using Parakeet and Qwen 3.5 models with adaptive policies. Their system achieved a +5.82 XCOMET-XL qualit…