AI news for Friday, 23 October 2026
Stories 61 to 120 of 120, most important first. Back to the first page →
-
AQL-Lite improves Arabic LLMs with curated data and partial-parameter training
Researchers introduced AQL-Lite, a framework designed to efficiently adapt multilingual large language models (LLMs) to Modern Standard Arabic (MSA). The method addresses challenges like noisy web data, dialectal…
1 source primary sourcearXiv cs.CL -
Google CEO Sundar Pichai warns AI computing shortages could last until 2027
Alphabet CEO Sundar Pichai told the Big Technology Podcast that the AI industry is already short of computing capacity and that the shortage could persist through 2027. He said Google’s current capacity reflects…
1 sourcecnbctv18.com -
OpenAI scales Azure Cosmos DB to serve 1 billion ChatGPT users
OpenAI has detailed how it scaled its internal storage platform, Habitat, to support over 1 billion users across 40 geographic regions. The system now manages more than 500 petabytes of data, with the largest single…
1 source primary sourceOpenAI -
Microsoft Teams adds AI phishing detection for external chats in November
Microsoft will roll out AI-powered phishing detection in Teams starting early November 2026, with general availability expected between early and mid-December. The feature analyzes conversation context to spot…
1 sourcebleepingcomputer.com -
IBM CEO says businesses prefer open-weight models to costly frontier AI
IBM chief executive Arvind Krishna told investors that many enterprises are shifting to open‑weight AI models because they can deliver roughly 90% of the performance of leading closed models at about 10% of the cost.…
1 sourcehotminute.co.uk -
Ex-DeepMind researcher says a one‑in‑three chance AI will kill humans in ten years
Alex Turner, a former DeepMind researcher who left Google in July after the company signed a Pentagon AI contract, warned at the NDTV World Summit that there is a one‑in‑three chance artificial intelligence could kill…
2 sourcesdailymail.comradaronline.com -
Texas Athletics and McCombs School launch Bevo Lab with Dell, Google Cloud, SHI
University of Texas Athletics and the Business of Sports Institute at the McCombs School of Business announced the creation of Bevo Lab, a sports‑intelligence platform that blends AI, data analytics and hands‑on…
1 sourceaol.com -
AgentionAI releases Qwen3.8-Flash-Next-Gyro-Coder for 24 GB GPUs
AgentionAI has published Qwen3.8-Flash-Next-Gyro-Coder, a quantized version of the Qwen3.8-Flash-Next model designed to run on a single consumer GPU with 24 GB of memory, such as an RTX 3090, 4090, or 7900 XTX. The…
1 source primary sourcehuggingface.co -
Razer Basilisk V4 Pro adds AI prompt button to gaming mouse
Razer has released the Basilisk V4 Pro, an update to its all-around gaming mouse line that introduces a dedicated AI prompt button. The mouse features Razer's Focus Pro 50K Gen-3 optical sensor, supporting up to 50,000…
1 sourceTom's Hardware -
Researchers use human-AI loop to break high-degree barrier in Lehmer's problem
A team of researchers has used a human-AI discovery loop to make progress on Lehmer's problem, a long-standing question in number theory from 1933. The problem asks whether the Mahler measure of an integer polynomial…
1 source primary sourcearXiv cs.AI -
Perplexity releases Decider for Mac to run pplx-decider-flash-v1.2 locally
Perplexity has made its 4 billion‑parameter decision model, pplx‑decider‑flash‑v1.2, runnable on a Mac without sending any data off‑device. The free Decider for Mac app installs a local server that mirrors the hosted…
1 sourcedocs.perplexity.ai -
LLM agents repeat failed moves in games, study finds
Researchers examined whether LLM agents repeat failed moves in board games when the harness logs the failure. They tested eight instruction‑tuned models ranging from 135 M to 1.7 B parameters on Sokoban, Connect Four,…
1 source primary sourcearXiv cs.AI -
Nvidia may cut engineering effort with AI systems at Siemens plant
Nvidia is rolling out a suite of AI tools at Siemens’ Electronics Factory in Erlangen, Germany, including digital twins, AI‑generated training data and AI agents. The partners say the AI‑driven photorealistic PCB…
1 sourcetechtarget.com -
TorchNeuron Space lists which Hub models run unmodified on AWS Trainium
A new Space called torch-neuronx model support lets users check whether any model on the Hub can run unmodified on AWS Trainium. The interface shows results for the most downloaded and trending models, indicating if…
1 source primary sourceHugging Face -
Infosys reports AI revenue at 10.5% of Q2 sales, up 29% quarter‑on‑quarter
Infosys posted Q2 revenue of $5,145 million for the quarter ended September 30, 2026, with a 21.0% operating margin. AI‑related revenue represented 10.5% of total sales and grew 29% from the prior quarter, according to…
1 sourceUnite.AI -
MIT SHASS marks 75 years, emphasizes human‑centered AI amid rapid change
The MIT School of Humanities, Arts, and Social Sciences celebrated its 75th anniversary with a two‑day symposium that highlighted the growing impact of artificial intelligence on teaching, research, and society.…
1 source primary sourceMIT News on AI -
Bold Security deploys on-device AI to guard enterprise data from shadow AI
Bold Security’s CEO Nati Hazut explains that the shift of work from browsers to local devices has created new data‑exfiltration channels that traditional network‑based security tools miss. By running small language…
1 sourceUnite.AI -
TikTok U.S. seeks four gigawatts of compute power by early 2030s
TikTok's U.S. entity, TikTok USDS Joint Venture, is looking to secure multiple gigawatts of computing capacity, with sources telling CNBC that it wants four gigawatts by the early 2030s. The joint venture runs…
1 sourceCNBC Technology -
Study measures handshake overhead in MCP, A2A, and ACP agent protocols
Researchers released the first empirical benchmark measuring the "handshake" overhead in three major AI agent communication protocols: MCP, A2A, and ACP. The study, published on arXiv, instruments every request to…
1 source primary sourcearXiv cs.AI -
Amazon scrubs plans to build four data centers in Loudoun County
Amazon scrapped plans to build four data centers on the former George Washington University campus in Loudoun County, Virginia. The 122‑acre site will instead be repurposed for mixed‑use and residential development.…
1 sourceTom's Hardware -
Apple unveils $299 HomeView smart home display with Siri AI
Apple introduced HomeView, a $299 7‑inch smart‑home hub that combines a display, speaker, FaceTime, Siri AI, and a camera. Preorders began on Oct 13 for the United States, Australia, Canada, India, New Zealand, and the…
1 sourceSmall Business Trends -
AI tools speed Retro game ports, raising preservation concerns
AI coding tools have turned the slow, painstaking process of decompiling classic video games into a rapid hobby. Projects like Halo in a web browser, Wind Waker on iPhone, and Super Smash Bros. in mixed reality have…
1 sourceThe Verge AI -
Wdowiak customizes NLLB for sicilian translation
Wdowiak rebuilt a small neural machine translation system for Sicilian by extracting parallel text from the bilingual journal Arba Sicula and porting the original tokenization to a Perl‑free implementation. The new…
1 source primary sourcearXiv cs.CL -
ARK Invest buys 38,110 Meta shares across three ETFs amid AI spending concerns
ARK Invest added 38,110 Meta Platforms (META) shares on Thursday, splitting the purchase among its ARK Innovation ETF (28,069 shares), ARK Next Generation Internet ETF (7,018 shares) and ARK Fintech Innovation ETF…
1 sourcefinance.yahoo.com -
Researchers introduce MemPABench to evaluate personal agent memory
A new benchmark called MemPABench has been released to test how personal AI agents remember and act on users' interaction preferences. The environment simulates four grounded users, each with 14 work‑ and…
1 source primary sourcearXiv cs.AI -
Martha Minow warns AI threatens democracy, suggests better tools
Martha Minow, a Harvard Law School professor, delivered the fall Compton Lecture at MIT, arguing that artificial intelligence is currently undermining democratic foundations while offering potential solutions if…
1 source primary sourceMIT News on AI -
TauIndianBankBench evaluates procedure compliance in indian banking assistants
The authors introduce TauIndianBankBench, a new benchmark designed to evaluate conversational agents that use language models in Indian retail banking. The benchmark contains 1,000 simulated customer interactions that…
1 source primary sourcearXiv cs.CL -
Tom Lee says tech selloff near bottom, recommends buying Nvidia and crypto
Fundstrat’s Tom Lee told CNBC that the recent tech selloff is close to bottoming and that investors should buy into the weakness, highlighting Nvidia, the Magnificent Seven, software, crypto, financials and cyclical…
1 sourcecoinpedia.org -
NVIDIA, AMD, Apple local AI hardware compared for inference speed
A new comparison evaluates local AI hardware from NVIDIA, AMD, and Apple, finding that the best choice depends on workflow compatibility rather than raw speed alone. The NVIDIA DGX Spark offers the safest software bet…
1 sourcetech.yahoo.com -
Microsoft explains why AI loses track in long conversations
In a podcast released on October 6, 2026, Microsoft Research researchers Chad Atalla and Jennifer Neville described "surprising failures" that appear when large language models are used in multi‑turn dialogues. They…
1 sourcenote.com -
Study finds six AI models act like two due to correlated errors
A new analysis published on Towards Data Science argues that ensembling multiple large language models yields fewer independent opinions than the number of models suggests. The author tested this hypothesis using six…
1 sourceTowards Data Science -
arXiv paper presents RoleFlow Self-Evolving Multi-Role agent system
The paper introduces RoleFlow, a large‑language‑model‑based multi‑agent framework that can modify its own collaboration structure. It does so by applying feedback‑driven atomic edits and creating new roles on demand,…
1 source primary sourcearXiv cs.AI -
arXiv paper proposes three-pillar framework for safe AI agents
The paper released on arXiv (2610.30445v1) outlines a governance model for autonomous AI agents that emphasizes transparency, accountability, and trustworthiness. It argues that existing human‑in‑the‑loop and…
1 source primary sourcearXiv cs.AI -
Microsoft may rely on security and custom chips to revive cloud margins
Analyst commentary on Microsoft Corp. (MSFT) notes a split market view: investors are losing confidence in the AI trade while enterprise AI adoption speeds up. The author believes Microsoft is positioning its security…
1 sourceseekingalpha.com -
arXiv study maps gaps in AI research for low-resource languages
A new paper on arXiv analyzes 1,158 open-access papers from 2020 to 2025 to measure how AI research serves low-resource languages. The study introduces the Linguistic Inclusion Framework (LIF), which evaluates…
1 source primary sourcearXiv cs.CL -
Researchers introduce Neural networks with differentiable Belief-Function layers
The authors present a deep evidential pipeline that replaces the standard Dempster–Shafer normaliser with Yager’s 1987 rule of combination. The new rule is bilinear and avoids the $(1-\kappa)^{-2}$ term that causes…
1 source primary sourcearXiv cs.AI -
ETH Zurich quadruped robot solves giant Rubik’s Cube using only its feet
ETH Zurich Robotics Systems Lab released a video showing a quadruped robot autonomously solving a giant Rubik’s Cube using only its feet. The robot follows a three‑stage framework called Reflex that learns visual…
1 sourceIEEE Spectrum Robotics -
Researchers find self‑rewarded moral argumentation makes language models decisive but pliable
A new arXiv paper investigates how rewarding language models for producing high‑quality moral arguments affects their judgments. The authors train a reinforcement‑learning policy, starting from a state‑of‑the‑art…
1 source primary sourcearXiv cs.CL -
Google adds AI‑powered Enhanced Protection to Messages for Android
Google is rolling out an optional Enhanced Protection mode in its Messages app for Android, adding AI‑driven scam detection and smarter link checks. The feature leverages Google’s cloud‑based Gemini models to analyse…
1 sourcedigitaltrends.com -
RAVEL CEO Denise Muyco says AI factories need orchestration, not just more GPUs
RAVEL co-founder and CEO Denise Muyco argues that the AI infrastructure challenge is not a shortage of GPUs but a systems-engineering problem of making distributed, heterogeneous compute work together efficiently. In…
1 sourceUnite.AI -
Opinion: AI model predicts cancer survival from selfies, but its usefulness is uncertain
Researchers at Mass General Brigham have built a suite of AI tools called FAHR‑Face. The facial‑age estimator, FAHR‑FaceAge, was trained on more than 40 million facial images, while the survival predictor,…
1 sourceThe Guardian AI -
Vibe coding with Opus 5.5 may trigger GTA 6 fade trade
The rise of vibe coding, where users describe software in plain language for AI to build, has intensified following the release of Anthropic's Opus 5.5. This development has caused a sell-off in video game stocks as…
1 sourceseekingalpha.com -
BELT reduces LLM reasoning tokens by 41% on Qwen3-1.7B
Researchers introduced BELT (BELief-based Termination), a method designed to stop Large Language Models from continuing to reason after they have already reached a correct answer. Current reasoning models often waste…
1 source primary sourcearXiv cs.AI -
Lighthouse chatbot to our knowledge launched for Civil Rights history research
Lighthouse is a new digital assistant that lets users explore the Civil Rights period of the United States using multimodal data from South Carolina. The authors present it as the first system of its kind. The…
1 source primary sourcearXiv cs.CL -
AI can speed up hardware prototype cycles for startups
AI is beginning to accelerate parts of the hardware build cycle, especially for startups that rely on rapid 3D‑printed prototypes. The 2026 Stanford AI Index reported that 88% of surveyed organisations used AI in at…
1 sourceUnite.AI -
Researchers unveil AI surrogate model to accelerate Canadian wildfire hazard mapping
Researchers from AI and wildfire‑modelling teams introduced a new surrogate‑modelling framework to speed up national hazard mapping in Canada. The framework standardises training, fine‑tuning and evaluation of AI…
1 source primary sourcearXiv cs.AI -
arXiv paper surveys agentic multimodal AI for 6G security
The arXiv paper titled “Secure Agentic AI and Multimodal Intelligence in 6G Communication Systems” surveys how next‑generation 6G networks could embed multimodal intelligence and autonomous control. It outlines the…
1 source primary sourcearXiv cs.AI -
Peach Tech starts company-wide Claude certification program for all employees
The program covers two certifications: CCAO-F (Claude Certified Associate – Foundations) for general business use and CCDV-F (Claude Certified Developer – Foundations) for API development. Peach Tech will cover all…
1 sourcenote.com -
EU AI Act guide outlines five-stage compliance process for general-purpose AI
Unite.AI has published a practical guide explaining the EU AI Act, a risk-based legal framework that assigns specific obligations to AI providers, deployers, importers, and distributors. The article argues that…
1 sourceUnite.AI -
Study compares CLI and MCP for AI agents, recommends hybrid architecture
A new arXiv paper evaluates two ways AI agents can talk to external systems: the traditional command‑line interface (CLI) and the Model Context Protocol (MCP), a JSON‑RPC style protocol. The authors assess the…
1 source primary sourcearXiv cs.AI -
Five free AI coding agents you can try in 2026
AI coding agents that let you explore repositories, edit files, run terminal commands, debug errors and execute tests are now reachable without paying for API credits. The guide highlights five options – OpenCode,…
1 sourceKDnuggets -
AWS shares CDK construct to fix SageMaker AI auto scaling updates
AWS published a technical guide addressing a common failure mode when updating Amazon SageMaker AI real-time endpoints using AWS Cloud Development Kit (CDK). The issue arises because SageMaker requires auto scaling…
1 source primary sourceAWS Machine Learning Blog -
Opinion: security teams should add a deterministic second opinion to AI scans
AI‑generated code is expanding far faster than security teams can keep up. GitHub logged 7.38 billion commits in September 2026, more than five times the total from a year earlier, and Cloud Security Alliance found…
1 sourceUnite.AI -
Companies use “AI experience” in job ads to signal tool familiarity, not model building
The guide explains that the phrase “AI experience” on job postings is a marketing term without a standard definition. It breaks the usage into four categories: familiarity with AI tools, awareness of machine‑learning…
1 sourceKDnuggets -
Designing AI for payroll and admissions calls for deterministic guardrails
The author describes building two AI‑enabled products—an Australian payroll compliance platform and a college admissions tool—where a single mistake can cost money or alter a student’s future. In the U.S., one in five…
1 sourceUnite.AI -
Claude guide explains AI basics and differences from ChatGPT and Gemini
A new educational guide from Search: Claude clarifies common misconceptions about artificial intelligence by distinguishing between the broad field of AI, machine learning, and specific products like ChatGPT, Claude,…
1 sourcenote.com -
Opinion: Comedian claims meta's Instagram does not stop AI joke theft
George Brett Olson, a stand‑up comedian who began performing in 2021, says an AI‑generated video reposted his joke about AI taking jobs and immigrants. He discovered the clip after a friend sent it, noting the AI…
1 sourcebusinessinsider.com -
Researchers introduce MemPlan for text-based planning with PDDL
A new framework called MemPlan bridges language models and symbolic planning for partially observable text environments. The system uses memory-conditioned PDDL planning to reconstruct problems dynamically from…
1 source primary sourcearXiv cs.CL -
Opinion: Google One AI Plus may offer better value than $20 AI Pro
The author compares Google’s One AI Pro plan, which costs $20 a month and provides 5 TB of cloud storage, with the lower‑priced AI Plus tier that offers 2 TB. Recent changes to the One plans added the Spark‑now‑Task…
1 sourceandroidpolice.com -
MIT historian says early AI engineers added randomness to mimic human creativity
Benjamin Lindquist, an MIT historian, explains that mid‑20th‑century engineers deliberately introduced randomness into algorithms to model the “messy” human mind. He traces the idea to figures like John Pierce, who…
1 source primary sourceMIT News on AI
