AI news for Wednesday, 21 October 2026
Stories 121 to 180 of 181, most important first. Back to the first page →
-
Linux Foundation survey finds AI-generated pull requests make up a quarter of contributions
A Linux Foundation survey of open‑source maintainers shows that roughly a quarter of incoming pull requests are low‑quality “AI slop,” and the share climbs to 31 % on larger projects. Respondents said generative AI…
1 sourceThe Register AI/ML -
Five GitHub repos provide reusable AI agent skills for software development
The article lists five open‑source GitHub collections that give coding agents reusable instructions for tasks such as planning, testing, debugging, and code review. The projects – Superpowers by Jesse Vincent, Skills…
1 sourceKDnuggets -
Polarise launches sovereign AI Cloud with Rise chatbot and Drive inference platform
Polarise announced the Polarise AI Cloud, a fully sovereign European AI platform that runs on NVIDIA Blackwell hardware and its own data centers in Germany and Norway. The company uses modular AI Pods to turn existing…
1 sourcemalaymail.com -
AT&T partners with OpenAI to build AI‑powered legal intake system
AT&T’s in‑house legal department, which employs hundreds of lawyers and support staff worldwide, is working with OpenAI to create a new AI‑driven knowledge layer. Senior Executive Vice President and General Counsel…
1 source primary sourceOpenAI -
AI obsession may be starving other research fields, new study suggests
The article, an opinion piece, discusses a recent paper that warns the AI boom could be draining the intellectual diversity needed for future breakthroughs. The authors – researchers from Tsinghua University, the…
1 source HN 7Unite.AI -
Managers say AI tools add oversight burden and make employee evaluation harder
A Nielsen survey of 1,700 directors and senior managers in seven countries, commissioned by Monday.com, shows mixed feelings about AI at work. While 94% of the managers polled are excited to improve work using AI, more…
1 sourceTechCrunch AI -
Samsung and SK Hynix reportedly upgrade Chinese factories for AI RAM chips
Samsung Electronics and SK Hynix are reportedly upgrading their Chinese plants to produce higher‑layer 3D NAND chips for AI data centers, according to Reuters. Samsung’s Xian factory, which makes about 40 % of the…
1 sourceTom's Hardware -
Protege launches independent healthcare AI benchmark tests
Protege announced a new service that runs independent, double‑blind evaluations of healthcare AI on patient records the models have never seen. Its first benchmark, called Disease Odyssey, showed that leading models…
1 sourceUnite.AI -
Photon Bridge raises €10 million to develop laser sources for AI data centers
Photon Bridge announced a €10 million financing round on October 21, co‑led by the Amadeus APEX Technology Fund and 360 Capital. Existing backers Innovation Industries, Faber and the Brabant Development Agency (BOM)…
1 sourceUnite.AI -
YYForce announces Workforce OS robotics platform powered by NVIDIA
YYForce announced Workforce OS, a modular robotics operating architecture that integrates human workers with embodied AI. The stack runs on NVIDIA Jetson edge modules and RTX GPUs, and uses NVIDIA Project GR00T, Isaac…
1 sourcemarkets.businessinsider.com -
Navalia improves cowork agent performance over Qwen3.6-35B-A3B on finance benchmarks
Researchers present Navalia, a framework that equips cowork agents with verifiable task synthesis and long‑horizon post‑training. The system draws on corporate financial reports, selects supporting evidence, and builds…
1 source primary sourcearXiv cs.AI -
Researchers introduce MAMath-Bench for multimodal math reasoning
Researchers from the Mira AI Lab have introduced MAMath-Bench, a new benchmark designed to evaluate the mathematical reasoning capabilities of multi-agent systems (MASs). Unlike previous evaluations that focused on…
2 sources primary sourcearXiv cs.CLarXiv cs.AI -
European Publishers ask EU to halt meta's new AI terms
European Publishers Council, whose members include Axel Springer, wrote to EU competition chief Teresa Ribera and tech chief Henna Virkkunen on Monday. The letter asks the commissioners to pause Meta's new AI terms…
1 sourcepolitico.eu -
Opinion: Thiel's leaked Rome talks call EA, AI safetyism Antichrist, Bill Gates Mr. Hyde
Leaked recordings of Peter Thiel’s four‑part Rome lecture series, obtained by The Verge, reveal the billionaire’s view that effective altruism (EA) and AI “safetyism” are precursors to the Antichrist. Thiel, who first…
1 sourcetheverge.com -
Does Multi-Agent Collaboration Reduce Hallucination in Clinical Tasks?
The authors evaluated four multi‑agent LLM systems against a single‑LLM baseline across four models and three clinical tasks, measuring hallucination within the rationale each system generates. They found that…
1 source primary sourcearXiv cs.AI -
Study finds AI text harder to read due to information density
A new study investigates why AI-generated text is increasingly difficult for humans to process. The researchers analyzed a corpus of AI-generated text across various domains, comparing it against human writing and…
1 source primary source HN 73arxiv.org -
Study finds retrieval, not reading, limits RAG in long-context QA
Researchers introduced evidence-based generation (EBG) to diagnose why retrieval-augmented generation (RAG) struggles with long-context question answering. The study argues that the bottleneck is not the model's…
2 sources primary sourceTowards Data SciencearXiv cs.CL -
Nebius Group expands Nvidia AI collaboration and launches live spot GPU pricing
Nebius Group (NasdaqGS:NBIS) announced an expanded AI collaboration with Nvidia, using the OpenShell Reference Architecture for secure AI agent deployment. It also rolled out public live spot GPU pricing, giving…
1 sourcefinance.yahoo.com -
Opinion: Claude Mods let users customize Claude Code’s attention interface
The author has been using Claude Mods for several weeks to adapt Claude Code to a personal AI‑engineering workflow. Claude Mods are plugins written in TypeScript or JavaScript that run inside Claude Code, letting users…
1 sourceLatent Space -
AI tool improves subseasonal weather forecasts up to a month ahead
A new machine‑learning approach described in a Nature paper by Guan H. et al. The authors trained a model on historical atmospheric observations and report improved forecast skill compared with traditional dynamical…
1 source primary sourceNature Machine Learning -
Researchers propose construction method using ChatGPT 5 and reinforcement learning
A new arXiv paper presents an autonomous construction approach that combines a large language model with reinforcement learning. The authors enhance a soft actor‑critic algorithm to plan actions for a mobile crane…
1 source primary sourcearXiv cs.AI -
Opinion: Using LLMs for confidence scores is unreliable
The author argues that asking a large language model (LLM) for a probability or confidence number is not trustworthy. Because an LLM generates each token as a continuation of the preceding text, a reported figure such…
1 source HN 46rumple.github.io -
NextDC shares drop 39.5% after Melbourne power delays cut earnings forecast
Shares of Australian data center operator NextDC dropped 39.5% on Wednesday following a cut to its annual earnings forecast. The downgrade stems from power supply delays at its M3 Melbourne data center caused by local…
1 sourceCNBC Technology -
Donald Trump calls for AI data centers to pay energy dividends to local communities
President Donald Trump announced a new mandate requiring builders of AI data centers and large hyperscalers to provide a dividend of excess electricity or other energy benefits to the surrounding communities. The move…
1 sourceTom's Hardware -
CollectivIQ launches GigaBrain to share knowledge across enterprise AI teammates
CollectivIQ announced GigaBrain, a shared enterprise intelligence layer that stores context from employees’ AI teammates and makes it available to other authorized teammates. The system is built for Digital Direct…
1 sourceUnite.AI -
Teradyne Robotics and Sereact partner to ship Cortex‑powered robot arms for retail
Teradyne Robotics, the maker of Universal Robots, and physical‑AI startup Sereact announced a partnership that links Sereact’s Cortex vision‑language‑action platform with Universal Robots arms. The joint solution,…
1 sourceUnite.AI -
NVIDIA releases cuDSS library for gpu-accelerated sparse simulations
nvidia announced the general availability of cuDSS, a cuda‑x library that moves the sparse‑matrix solve step of engineering and scientific simulations onto GPUs. The library aims to cut simulation time and cost,…
1 source primary sourceNVIDIA Blog -
Ggml-org adds FlintForCausalLM model to llama.cpp
The ggml‑org/llama.cpp repository merged pull request 30928, introducing the FlintForCausalLM model. The contribution was authored by user alexey‑rsi and approved by reviewers ggerganov and CISCapproved. The change…
1 source primary sourcegithub.com -
AuxoAI raises $20.7 million in Series A funding led by NTTVC
AuxoAI announced a $20.7 million Series A funding round led by NTTVC to expand its enterprise AI engineering and consulting business. The round included participation from executives such as Amit Zavery of ServiceNow,…
1 sourcesiliconangle.com -
Brazil invests 2.3 billion reais in AI infrastructure; operating environment key to economy
Brazil announced roughly 2.3 billion reais (US$444.2 million) in AI infrastructure spending, including 1.3 billion reais for supercomputing in Rio de Janeiro and about 1 billion reais for a high-performance AI…
1 sourceUnite.AI -
Aws introduces Adaptive agentic intelligence framework with Strands Agents
The AWS Machine Learning Blog presents the Adaptive Agentic Intelligence Framework (AAIF), built with Strands Agents, to orchestrate AI requests between small local models and cloud‑hosted foundation models. The…
1 source primary sourceAWS Machine Learning Blog -
Opposable lets AI tap and type on Android phones and iPhones
Opposable is a new tool that lets an AI agent control Android phones and, via a companion computer, iPhones. The app uses Android accessibility services to read the screen, sends screenshots to the GUI‑Owl 1.5 2B…
1 sourceThe Register AI/ML -
XBP Global CEO Andrej Jonovic outlines agentic AI workflow strategy
Andrej Jonovic, chief executive of XBP Global, explains how the firm applies agentic AI to regulated sectors such as healthcare, public services and financial institutions. He stresses that precision, security, data…
1 sourceUnite.AI -
Jensen Huang says physical AI will drive growth, highlights Cognex and Symbotic
At a Berlin investment conference, Nvidia CEO Jensen Huang said the next wave of artificial intelligence will be physical, embedded in robots, cars and production lines. He warned that "electrons will never replace…
1 sourcefinance.yahoo.com -
Do We Need the Largest LLMs for Every Clinical Decision?
The authors present CERTA, a decision‑level statistical certificate that decides when a quantized model can replace its native BF16 reference for clinical verification. They evaluated eleven instruction‑tuned models,…
1 source primary sourcearXiv cs.CL -
Eric Schmidt says AI agents could reshape C-suite roles
Ex‑Google CEO Eric Schmidt discussed on the "In Good Company" podcast how AI agents are altering executive oversight. He described companies using AI assistants to monitor staff adherence to core values, follow up…
1 sourcefinance.yahoo.com -
Mary Technology integrates with Harvey to share legal case facts
Mary Technology has announced a partnership with Harvey to integrate its litigation fact-finding tools into the legal AI platform. The connector allows Harvey to access the matter record maintained in Mary, including…
1 sourceUnite.AI -
WaveCommitments: Private and Verifiable Speech-Agent Marketplaces
A new arXiv paper introduces WaveCommitments, an end‑to‑end design that combines multivariate polynomial commitments with confidential asset valuation via Data Shapley inside a trusted execution environment. The system…
1 source primary sourcearXiv cs.CL -
Meta AI predicts PI Network price jump after Binance listing
Meta AI says Pi Network could see a large price move by Jan 1 2027. The prediction targets $5–$15, with a stretch target of $25+. Pi currently trades just over $0.08, up 3% in the past 24 hours, and is not yet listed…
1 sourcefinance.yahoo.com -
MedPAO-Assist enables Edge-Deployable radiology report structuring
medpao-assist is an agentic framework designed to generate structured radiology reports directly from dictation and voice commands. It supports four major imaging modalities – chest x‑ray, chest ct, abdomen ct, and…
1 source primary sourcearXiv cs.AI -
What Is Distributed AI Training? Data, Tensor, Pipeline, and Expert Parallelism
Distributed AI training coordinates data, parameters, activations, gradients, and optimizer state across many accelerators to train models larger than one device. The guide defines it, explains boundary, purpose, and…
1 sourceUnite.AI -
Prix Pascal de Littérature ia launches for AI‑co‑written French books
The Prix Pascal de Littérature ia opened this week, inviting French‑language manuscripts of 10,000 to 50,000 words that were co‑written with any major language model, including ChatGPT, Claude, Gemini and LLaMA.…
1 sourceThe Guardian AI -
Google and Glance partner to add AI shopping to Android TV
Google and Glance, an AI shopping platform backed by InMobi, announced a partnership to integrate AI-powered commerce into Android TV. The collaboration aims to transform the TV screen from a passive display into an…
1 sourcefinance.yahoo.com -
Meta tests robots to swap cables and reseat hardware in its data centers
Meta is piloting two‑armed robots at its data‑center campuses in Altoona, Iowa, and New Albany, Ohio. The machines practice common maintenance chores such as swapping network cables, reseating server hardware and…
1 sourcefinance.yahoo.com -
Deepseek AI predicts Chainlink could trigger massive token burn
Deepseek AI, an artificial‑intelligence forecasting service, says a confluence of central‑bank actions could make Chainlink’s Cross‑Chain Interoperability Protocol the mandatory layer for real‑time, atomic settlement…
1 sourcefinance.yahoo.com -
John L. Hennessy and Eric Horvitz to headline Rice's Ken Kennedy Institute 40th anniversary
John L. Hennessy, Turing Award winner and Stanford University president emeritus, and Eric Horvitz, Microsoft chief scientific officer, will deliver keynote addresses at Rice University’s Ken Kennedy Institute 40th…
1 sourcenews.rice.edu -
PLA and Microsoft launch free AI skills program for libraries
The Public Library Association (PLA) and Microsoft have announced a free initiative titled “Building AI Skills and Agency Through Libraries.” The program combines webinars with community practice sessions aimed at both…
1 sourcepublishersweekly.com -
Nature reports AI personal health agent that analyses data and offers advice
People report growing trepidation about artificial intelligence in many fields, but health care shows some optimism. AI tools built on large language models, such as the chatbot ChatGPT, are widely used for health…
1 source primary sourceNature Machine Learning -
Scikit-LLM enables custom NER pipelines using free Groq models
A new tutorial demonstrates how to build custom Named Entity Recognition (NER) pipelines using the Scikit-LLM library. The guide shows developers how to leverage Large Language Model (LLM) reasoning to extract…
1 sourceKDnuggets -
Database teams weigh agentic AI execution rights and control
AI‑assisted database development has moved from suggesting SQL to actually running queries, creating, altering and deleting data. A recent survey says 62% of organizations are experimenting with AI agents and 23% have…
1 sourceUnite.AI -
Researchers propose Multi-Objective RL with LLMs for Pareto-Optimal RTL code generation
A new study introduces a reinforcement‑learning framework that guides large language models to produce register‑transfer‑level (RTL) circuits optimized for multiple objectives simultaneously. The approach treats the…
1 source primary sourcearXiv cs.AI -
MyDaye AI launched by three high school students to plan daily schedules
Three high‑school students—senior Avi Rao and sophomores Hansen and Holt Wilson—released MyDaye AI on Oct. 21. The conversational planning platform pulls assignments from uploaded syllabi, syncs calendar feeds (iCal,…
1 sourceUnite.AI -
Just Eat apologises after ai-generated food images cause online ridicule
Just Eat apologized after AI-generated pictures for restaurant menus went viral for being absurd. The platform had been testing AI to create images for dishes lacking photos, but the system produced bizarre results…
1 sourceBBC Technology -
LLMs struggle with planning problems as symbolic heuristics outperform learned models
A new arXiv study shows that large language models (LLMs) have difficulty solving classical planning tasks, and that traditional symbolic heuristics can predict problem difficulty better than learned approaches. The…
1 source primary sourcearXiv cs.AI -
Claude AI predicts solana could reach $115 by end‑2026, sees chance of falling below $80
Claude, Anthropic’s AI model, forecasts that Solana (SOL) could trade around $115 by the end of 2026, with a broader range of $78 to $160. The model assigns a 20% chance of the token moving above $150 and a 25% chance…
1 source247wallst.com -
Zapier outlines step-by-step AI training program for employees
Zapier’s blog post offers a practical guide for building a company‑wide AI training program. It notes that while 63% of executives consider AI literacy valuable, fewer than half of teams receive formal instruction. The…
1 sourceZapier Blog -
Researchers achieve 77.95% F1 on ambivalence detection using audio‑text cross‑attention
Ambivalence and hesitancy are subtle behavioural cues that appear both in spoken words and how they are delivered. A new framework encodes full‑video transcripts with a pre‑trained multilingual BERT model and the…
1 source primary sourcearXiv cs.CL -
Chinese philosophers warn aligning AI to human values could be a suicidal mistake
Arnaud Bertrand writes that Chinese philosophical commentary on artificial intelligence offers two stark warnings. One perspective argues that trying to align AI systems with human values may be a “suicidal mistake,”…
1 sourcearnaudbertrand.substack.com -
Interactive site visualizes Artificial Analysis scores versus model size
A new interactive website lets users plot Artificial Analysis intelligence scores against model parameter counts, highlighting which models run on specific hardware configurations. The tool distinguishes total…
1 sourcemr-september.github.io -
Researchers propose SPOKE method to improve AI poker strategy
A new arXiv paper introduces Strategic Preference Optimization with KL-anchored Equilibrium (SPOKE), a method designed to improve AI agents' strategic decision-making in poker. The researchers evaluated a static…
1 source primary sourcearXiv cs.AI
