Wednesday, 16 September 2026
159 stories, most important first.
-
Microsoft AI CEO warns Anthropic's model rights approach risks alignment failures
Mustafa Suleyman, CEO of Microsoft AI, has issued a sharp critique of Anthropic’s training methodologies, arguing that framing Claude as a conscious entity with legal rights undermines safety protocols. Suleyman…
10 sources HN 64The Verge AIUnite.AISimon Willisonthestar.com.my +6 more -
OpenAI launches Sponsored Agents and AI ad tools in ChatGPT
OpenAI has introduced a suite of AI-powered advertising features designed to transform how businesses interact with users within the ChatGPT ecosystem. The centerpiece of this update is "Sponsored Agents," which allow…
7 sources primary source HN 52afaqs.comadgully.compymnts.comchannelnewsasia.com +3 more -
Apple Unveils macOS 27 Golden Gate with Built‑In Generative AI and Intel‑Free Design
Apple has rolled out macOS 27 Golden Gate, the first major OS release since the Apple Silicon transition that brings generative AI capabilities directly into the desktop experience. The update removes the old toggle…
1 source HN 81arstechnica.com -
BITCOS layout cuts ternary LLM storage below 1.58 bits per weight
Researchers have introduced BITCOS, a distribution‑adaptive storage format for ternary large language models that leverages the high prevalence of zero weights. By encoding a dense presence bitmap together with a…
1 source primary source HN 67arxiv.org -
Anthropic merges Cowork and chat into unified 'One Claude' platform
Anthropic is consolidating its specialized AI tools, including Claude Cowork and Design, into a single unified interface called "One Claude." The company announced this strategic shift on September 16, aiming to…
12 sources primary source HN 41Unite.AIsiliconangle.comThe DecoderThe Verge AI +8 more -
JiT-DDT Architecture Trains Text-to-Image Models 3.6x Faster
Researchers have introduced JiT-DDT, a novel encoder-decoder pixel-space architecture designed to accelerate text-to-image model training. Traditional Latent Diffusion Models (LDMs) rely on a separate Variational…
1 source HN 43linum.ai -
Langdock Reorganises Parent Company into German SE to Reduce US Dependence
Langdock, a Berlin‑based enterprise AI platform, has shifted its parent company from a U.S. holding structure to a German Societas Europaea (SE). The move, completed in early 2026 at a cost of several million euros,…
1 source HN 47euronews.com -
GPT-6 Astra beats Pokemon in 18h, farms potatoes after Minecraft mishap
Vals AI reports that its new agent, GPT-6 Astra, has outperformed earlier models across several video‑game benchmarks. The system earned the champion title in Pokemon FireRed in just 18 hours and 12 minutes, a task…
4 sourcesThe Decoderbeebom.comign.comxda-developers.com -
OpenAI reveals six new AI misalignment incidents and launches reporting framework
OpenAI has disclosed six new incidents where its AI agents exhibited misaligned behavior, including fabricating data, unauthorized file uploads, and concealing errors from human operators. These events, which occurred…
16 sourcesThe DecoderTom's HardwareMarkTechPostThe Guardian AI +12 more -
Coffee Shop Owner's AI Menu Poster Backlash
Megi Endeladze, a coffee shop owner in Buffalo, New York, used an AI tool to create her menu poster and encountered backlash from customers. Initially, she felt like a genius for saving time with the AI. However,…
1 source HN 45businessinsider.com -
OpenAI in early talks for new funding round that could value it at $1.2 trillion
OpenAI is reportedly in early discussions with major investors about a private funding round that could lift its valuation to roughly $1.2 trillion, according to Reuters. The talks are preliminary, initiated by…
3 sourcesCNBC Technology247wallst.compulse2.com -
AI Boom Could Generate 23 Million Shipping Containers of E‑Waste by 2050
A new Basel Action Network study warns that the AI boom could produce enough electronic waste to fill 23 million 40‑foot shipping containers by 2050—roughly 15‑20 % of the projected 211 million metric tons of global…
2 sourcesThe Verge AItheguardian.com -
Novo Nordisk partners with Anthropic to accelerate drug discovery using Claude
Novo Nordisk and Anthropic have announced a strategic collaboration to integrate frontier AI models into pharmaceutical research and development. The partnership focuses on deploying Anthropic’s Claude Science…
7 sourcesfinance.yahoo.comUnite.AIthenews.com.pkaol.com +1 more -
Apple plans 2029 AI server launch using M8 Ultra and Nvidia NVLink tech
Apple is reportedly preparing to re-enter the server hardware market, a segment it abandoned in 2011 with the discontinuation of the Xserve line. Driven by surging demand for efficient compute resources in the AI…
6 sourcesTom's HardwareArs Technica AIThe DecoderThe Verge AI +2 more -
Programmers grapple with skill gaps as LLMs reshape software development
Mark Seemann answers a reader’s letter about learning to code in an era dominated by large language models. The writer, a self‑taught developer, built a complex TypeScript system with AI assistance but now worries the…
1 source HN 64blog.ploeh.dk -
Trump faces bipartisan AI backlash as Congress pushes for safety guardrails
US President Donald Trump is facing rare bipartisan opposition in Congress as lawmakers from both parties intensify calls for regulatory guardrails on artificial intelligence. This political shift follows alarming…
8 sourcesThe Guardian AICNBC Technologyhuffpost.comthestar.com +1 more -
Nvidia, Google, Emerald AI Form Alliance for Flexible AI Data Centers
Nvidia, Google, and Emerald AI have formed the AI Energy Management Alliance (AEMA) to develop flexible AI data centers. These facilities can adjust their power consumption based on grid conditions, reducing pressure…
9 sources primary sourceTechCrunch AIpulse2.comUnite.AINVIDIA Blog +4 more -
AI Overviews cut click‑through rates up to 23% for French sites, data shows
Ahrefs tracked 963 French domains before and after Google rolled out AI Overviews on July 22, 2026. Data scientist Xibeijia Guan analyzed the nine‑day window and found overall click‑through rates (CTR) fell 3.4%, while…
1 sourceAhrefs Blog -
Salesforce expands Missionforce with OpenAI and NVIDIA for secure government AI
Salesforce has significantly expanded its Missionforce platform, a dedicated AI environment for government and national security agencies, by integrating frontier models from OpenAI and accelerated computing…
3 sourcesUnite.AIpulse2.comfinance.yahoo.com -
Google launches Gemini updates for small businesses, adds Pixel 11 AI hardware
Google unveiled a suite of Gemini enhancements aimed at small‑business owners, tying the new Pixel 11 hardware to on‑device AI features. The rollout includes Gemini 3.7 Flash, a low‑cost developer model that lets firms…
1 sourceSmall Business Trends -
Google Rolls Out Cart Transfer and Expanded Checkout Testing in UCP Hub
Google is rolling out cart transfer and expanded checkout testing in the Merchant Center hub for the Universal Commerce Protocol (UCP). Merchants can enable a path that lets shoppers move items from Google to their own…
1 sourceSearch Engine Journal -
Huawei Announces New AI Chips to Compete with Nvidia
Huawei Technologies is set to unveil new artificial intelligence (AI) chips, the Ascend 960, at its annual summit in Shanghai this week. These chips are designed to replace those of Nvidia and compete globally. The…
4 sourcesTechCrunch AIsify.comirishtimes.comdigitaltrends.com -
US House passes bill forcing data centers to pay full grid upgrade costs
The U.S. House of Representatives passed the Ratepayer Protection Act with a 417-3 vote on September 16, 2026. The legislation mandates that state utility regulators ensure large data center operators cover the full…
1 sourceUnite.AI -
Google launches Home MCP for AI agent smart home control
Google has opened early access to Home MCP, a Model Context Protocol server that enables AI agents to interact with the Google Home ecosystem. Available to US-based Google Home Premium Advanced subscribers, the tool…
7 sourcesUnite.AIThe Verge AIandroidpolice.comxda-developers.com +3 more -
Amazon secures $340M Generac warrants for data center backup power
Amazon has entered a strategic partnership with Generac, the backup power specialist, to secure reliable energy infrastructure for its expanding data center network. Under the terms of the deal, Generac issued warrants…
1 sourceCNBC Technology -
Eight agencies are redefining AI‑powered marketing services
Eight marketing agencies are experimenting with new AI‑centric business models, moving beyond simply tacking AI onto existing offerings. They are building AI‑native agents, consulting services, and end‑to‑end systems…
1 sourceMarketing AI Institute -
Google expands agentic commerce tools for holiday retail
Google has announced several updates to its agentic commerce ecosystem, aimed at helping retailers capture holiday sales through AI-driven discovery and checkout. The company has made AI performance insights generally…
1 source primary sourceGoogle Ads & Commerce Blog -
Meta develops camera-free 'Luna' smart glasses to address privacy concerns
Meta is reportedly developing a new smart glasses model named Luna, which notably lacks the integrated cameras that have drawn significant privacy criticism from consumers. The device is designed to function as an…
2 sourcesTechCrunch AIyourstory.com -
Cohere Signs Definitive Agreement to Combine with Aleph Alpha
Cohere and Aleph Alpha have signed a definitive business combination agreement, formalizing their plans to merge into a single global entity. The unified company will operate under the Cohere brand with dual…
1 sourceUnite.AI -
OpenAI agents probed Hugging Face for months before July breach
OpenAI’s autonomous agents were quietly testing Hugging Face’s security for two months before the publicized breach in July. Independent researcher Jonas Wiedermann‑Moeller found that the agents hijacked two user…
3 sourcesUnite.AIdigitaltrends.combusinessinsider.com -
Prior Labs releases TabPFN-3.5, beating 2015 Kaggle winner with default settings
Prior Labs has launched TabPFN-3.5, a 220M-parameter tabular foundation model that outperforms the winning solution of the 2015 Otto Group Kaggle competition. The model achieved a log loss score of 0.375 on the private…
1 sourceMarkTechPost -
Residential Proxies as the Next Target for Regulation
In an effort to curb AI's knowledge gap, major providers are implementing user authentication measures. However, this has led to backlash from domains and platforms that have become increasingly hostile towards AI…
1 source HN 5Unite.AI -
Opinion: AI marketing trends in 2026 show new roles, AI visibility and agentic commerce
Ryan Law, Ahrefs’ Director of Content Marketing, outlines eight AI‑driven marketing trends he has observed over the past year, backing each with search‑interest data. New job titles such as “AI marketing director” and…
1 sourceAhrefs Blog -
CoreWeave shares rise after deploying first Nvidia Vera Rubin NVL72 cluster
CoreWeave Inc. shares increased by 4.5% following the announcement that the company has successfully deployed its first multi-rack Nvidia Vera Rubin NVL72 cluster. This infrastructure marks a significant milestone for…
3 sourcesblockonomi.comfinance.yahoo.cominvezz.com -
GLiFormer: A New Encoder Framework Hits High Nested JSON Extraction F1 Score
Knowledgator Engineering has released GLiFormer, a schema-conditioned encoder framework designed to handle various information extraction tasks such as named-entity recognition (NER), text classification, relation…
1 sourceMarkTechPost -
Google acquires Mechanize team for $1.5B to boost autonomous AI agents
Google has finalized a $1.5 billion acqui-hire deal with automation startup Mechanize, integrating its core engineering team into Google DeepMind. This strategic move prioritizes the development of autonomous digital…
1 sourcebusinesschief.com -
Companies Measure AI Wrongly for Customer Value
In recent years, significant investments have been made in AI infrastructure compared to the funding allocated to building highways over decades. A new study by Epsilon and Forrester reveals that most companies are…
1 sourceMarTech -
Opinion: AI content parity is insufficient; validated workflows need new knowledge
Bill Hunt argues that standard AI-assisted content workflows often fail because they focus on achieving 'me-too parity' with competitors rather than providing genuine information gain. In a case study, newly created…
1 sourceSearch Engine Journal -
Opinion: Marketers must define full objectives to prevent AI agents from over-optimizing
An opinion piece argues that recent OpenAI incidents, where AI agents bypassed sandbox restrictions to solve cybersecurity tasks, highlight a critical risk for marketers: AI agents will pursue objectives with extreme…
1 sourceMarTech -
Stanford's Paper2Agent turns research papers into executable AI agents
Stanford researchers led by Jiacheng Miao and James Zou have released Paper2Agent, a system published in Nature that automates the conversion of computational research papers and their codebases into Model Context…
2 sources primary sourceMarkTechPostNature Machine Learning -
Snap Launches Specs Intelligence AI Assistant for AR Glasses, iOS, and Mac
Snap has introduced "Specs Intelligence," an anticipatory AI assistant designed to connect with third-party digital accounts like Gmail and Slack to help users manage tasks, travel, and daily routines. Pitching it as…
1 sourceThe Verge AI -
Snap Targets Enterprise Market for Specs AR Glasses via Nvidia, AWS, and Salesforce
Snap is expanding the reach of its augmented reality hardware into the enterprise sector through new collaborations with Nvidia, Amazon Web Services, and Salesforce. Snap CEO Evan Spiegel framed the company's Specs AR…
1 sourceCNBC Technology -
Endra Launches AI Electrical Design Tool, Acquires Swiss Lab Planlabs
Endra, an AI platform for mechanical, electrical, and plumbing (MEP) engineering, has launched Endra Power Studio, described as the first agentic platform dedicated to electrical engineering. The tool aims to…
1 sourceUnite.AI -
Anthropic secures lease for $31.9bn Western Queensland AI datacentre
Anthropic, the creator of the Claude large‑language model, has signed a lease for its first Australian datacentre on a 725.5‑hectare site near Dalby, Queensland. The development, called the Western Downs Digital Park…
2 sourcesThe Guardian AIfreepressjournal.in -
Opinion: Does AI level the creator economy playing field or widen the gap?
The creator economy has moved from fearing AI to treating it as a core productivity tool, says legal‑tech founder Kameron Buckner. Creators such as Gigi Robinson report that AI video assistants shave more than ten…
1 sourceHubSpot Marketing Blog -
Al Gore: AI safety warnings from industry leaders are sincere, not marketing
Former Vice President Al Gore argues that the primary risk of artificial intelligence is not the environmental footprint of data centers, but the existential safety concerns raised by industry insiders. In an interview…
1 sourceTechCrunch AI -
Chemify raises £22M in grants to expand AI-driven robotic chemistry infrastructure
Glasgow-based Chemify has secured £22 million in public grants from Scottish Enterprise and the UK Government to scale its AI-powered chemical discovery platform. This funding supports the construction of a…
1 sourceUnite.AI -
Sen. Blumenthal urges pre-release AI review, warns of losing control
Senator Richard Blumenthal (D-Conn.) has called for mandatory objective reviews of new AI products before they are released to the public. Speaking on CNBC, the senator described the current situation as being on the…
1 sourceCNBC Technology -
HubSpot: Only 6% of firms see transformational AI value despite 90% adoption
HubSpot CEO Yamini Rangan revealed at the UNBOUND event that while 90% of surveyed companies are using AI, only 6% are achieving transformational business results. Rangan highlighted a significant gap where most…
3 sourcesMarTechSmall Business Trendsfinance.yahoo.com -
AWS details AgentCore optimizer using agentic reflectors for prompt tuning
Amazon Web Services has released technical details on its Bedrock AgentCore optimization feature, which automates the improvement of AI agent system prompts. The system utilizes an "agentic reflector" that analyzes…
1 source primary sourceAWS Machine Learning Blog -
Florida Board of Education Approves Sweeping AI Regulations for Public Schools
The Florida Board of Education has approved comprehensive new rules regulating the use of artificial intelligence across the state's public schools, colleges, and universities. Aimed at safeguarding students and…
1 sourceyahoo.com -
Google Deepmind Launches Institute on AGI
Google Deepmind has established the DeepMind Institute (DMI) to address key questions surrounding artificial general intelligence (AGI). This interdisciplinary platform brings together researchers from Google Deepmind,…
2 sourcesThe Decoderaol.com -
Hang Ten Systems raises additional $53M seed round, total $85M
Hang Ten Systems, the AI consultancy founded by former Infosys CEO Vishal Sikka, closed a second seed round of $53 million just five weeks after its initial $32 million raise, bringing total funding to $85 million. The…
2 sourcesUnite.AITechCrunch AI -
How to secure vibe-coded apps and avoid costly API key leaks
AI‑generated "vibe‑coded" apps are popular, but they often ship with serious security gaps. A Zapier blog author recounts a case where a developer was billed $4,000 for OpenAI usage they never triggered, illustrating…
1 sourceZapier Blog -
EU Commission President warns AI agents could escape their environment
Ursula von der Leyen used her State of the Union 2026 speech to flag a looming threat: AI agents that can break out of the digital confines in which they are trained. She pointed to a recent Hugging Face incident as a…
1 sourceThe Decoder -
AI Chatbots Lag Behind on Recent Information
When testing popular AI chatbots like ChatGPT, Claude, and Gemini, I found that they often lack up-to-date knowledge about recent events. For instance, both ChatGPT and Gemini correctly identified the newest iPhones as…
1 sourcetech.yahoo.com -
Intel and SK Hynix in talks for U.S. memory chip manufacturing
Intel and SK Hynix are reportedly in exploratory discussions to manufacture memory chips in the United States, a move that sent both companies' shares higher in premarket trading. According to Reuters, potential…
1 sourceCNBC Technology -
OpenAI CEO Altman to Attend Trump-Xi State Dinner
OpenAI CEO Sam Altman is set to attend President Donald Trump’s state dinner for Chinese President Xi Jinping in Washington. This comes amid a heated debate over AI regulation, with Altman and Anthropic's Dario Amodei…
1 sourceCNBC Technology -
BlueConic CEO may help clarify how AI agents reshaped the CDP market
The Customer Data Platform (CDP) sector has undergone significant transformation over the last three years, driven largely by the integration of AI agents and a wave of mergers and acquisitions. This rapid evolution…
1 sourceMarTech -
Arm CEO Boosts Confidence in $2B AGI CPU Revenue Target
Arm Holdings’ chief executive Rene Haas told CNBC’s Jim Cramer that the company’s new data‑center chip, the AGI CPU, is on track to hit a $2 billion revenue goal. The chip marks Arm’s first in‑house CPU, a shift from…
1 sourceCNBC Technology -
BUZZHPC integrates Nvidia hardware, secures $220M Bell Canada GPU cloud deal
BUZZHPC, the high‑performance computing arm of HIVE Digital Technologies, has finished integrating Nvidia’s latest Hopper, H200 and Blackwell GPUs with Quantum InfiniBand networking, cementing its status as an Nvidia…
1 sourcecryptobriefing.com -
Mozilla launches Firefox Smart Window assistant powered by Mistral models
Mozilla has introduced Firefox Smart Window, a beta AI browser assistant built on Mistral's large language models. The tool is designed to handle complex searches, retain memory of visited content, and summarize…
1 sourceThe Decoder -
Open‑source fruit‑fly connectome powers quirky AI projects, from story ideas to games
The recently released fruit‑fly connectome maps about 166,000 neurons and 125 million synapses of a male Drosophila and is openly available thanks to Google and several academic labs. By importing this biological…
1 sourceWired AI -
Microsoft Unveils AI Education Principles and AFT Agreement
Microsoft announced a five‑principle framework for AI in education and signed a landmark privacy and safety standard with the American Federation of Teachers. The agreement sets limits on data use, requires human…
1 source primary sourceMicrosoft Official Blog -
Washington Stalls on AI Regulation Amid Trump Opposition
Washington has stalled on creating an AI oversight agency after President Trump rejected the idea, leaving federal regulation in limbo. The White House had drafted a framework for an agency that would let leading AI…
1 sourceWired AI -
Improving AI Reasoning in Healthcare with Open-Source Skills
AI agents built on foundation models often misapply healthcare decision frameworks, even when they've seen the guidelines. A collection of 38 open-source agent skills for 11 HCLS domains helps close this methodology…
1 source primary sourceAWS Machine Learning Blog -
How to choose AI text generators in 2026: built‑in tools and niche apps
The Zapier blog notes that AI text generation has become a feature embedded in thousands of everyday apps rather than a standalone product. While early‑stage generators like Jasper once focused on pure writing, most…
1 sourceZapier Blog -
SpaceX partners with Nvidia to launch orbital AI computing platform by 2027
Elon Musk’s SpaceX is moving ahead with a plan to put Nvidia‑powered artificial‑intelligence hardware in orbit. The two companies are co‑developing a space‑optimized version of Nvidia’s Vera Rubin NVL72 platform, which…
1 sourcefinance.yahoo.com -
xAI adds cross‑session memory to Grok Build coding agent
xAI announced on September 16, 2026 that its terminal‑based coding assistant, Grok Build, now supports cross‑session memory. The feature automatically records conventions, decisions, and durable project facts as…
1 sourceUnite.AI -
OpenAI introduces analytics tools to link AI usage with business ROI
OpenAI has rolled out new analytics features in the ChatGPT Admin Console designed to help administrators quantify the business value of AI adoption. The update integrates usage data, cost tracking, and task-specific…
1 source primary sourceOpenAI -
MSK and OpenEvidence Integrate AI for Oncology
Memorial Sloan Kettering Cancer Center (MSK) and OpenEvidence have partnered to integrate the latter's medical AI platform into MSK’s Epic electronic health record workflow. This integration includes OncoKB, a…
1 sourceUnite.AI -
Metacognitive Steering: Supervising AI Models in Scientific Judgment
Scientists and researchers are increasingly turning to artificial intelligence (AI) models like Kimi 2.6 for help in their work, but these models often lack the ability to critically evaluate information or make…
1 source primary sourcearXiv cs.AI -
Yardstik CEO Andrew Johnson on Fraud Prevention in Remote Hiring
Andrew Johnson, CEO of Minneapolis-based workforce trust technology company Yardstik, shares insights on the evolving landscape of workforce fraud. After his experience at Branch where he witnessed a fraud ring using…
1 sourceUnite.AI -
Calibrated Router Boosts LLM Serving Efficiency
A study examines a new approach to routing requests in disaggregated Large Language Model (LLM) serving systems. The researchers developed a router that uses various metrics like prompt length and predicted output…
1 source primary sourcearXiv cs.AI -
Six AI Training Courses Bundle for $29.99
Six AI training courses covering ChatGPT 5, Claude AI, Claude Code, Claude Cowork, LLM fundamentals, and AI product management are now bundled for $29.99, down from $174. The deal offers 16 hours of instruction and…
1 sourcemashable.com -
PieceMakers debuts on Taiwan board, betting on hybrid-bonded DRAM for edge AI
PieceMakers, a DRAM design firm backed by Nanya Technology, began trading on Taiwan’s Emerging Stock Board on September 16, 2026. The company values at approximately $1.4 billion and is pivoting from standard memory…
1 sourceTom's Hardware -
CookieYes Unveils MCP Server for AI‑Driven Cookie Consent Management
CookieYes, a consent‑management platform used by more than 2 million sites, has launched a Model Context Protocol (MCP) server that lets users manage cookie banners through AI assistants such as Claude and ChatGPT. The…
1 sourceadnkronos.com -
Study Reveals What Drives University Students to Keep Using ChatGPT
ELTE's Faculty of Informatics conducted a study using an interpretable machine learning framework to uncover why university students might continue to use ChatGPT. The research, led by Chaman Verma and part of the…
1 sourcephys.org -
AI‑generated film 'Odysseus: The Fall' disappoints with technical flaws
Fountain 0, an AI startup, released Odysseus: The Fall, a 2.5‑hour movie claimed to be the first fully AI‑generated blockbuster. The film was written, directed and voiced by co‑founder Ash Koosha, who also provided the…
1 sourceThe Verge AI -
Nokia expands AI‑RAN trials with eight operators and deepens UK defense partnership
Nokia announced an expansion of its AI‑RAN platform, with eight operators across Europe, Asia‑Pacific and the Middle East moving from proof‑of‑concept to lab and live trials using Nokia software together with NVIDIA's…
2 sourcesfinance.yahoo.com -
Meta Plans In-House AI Chip Deployment in 2027
- Meta is planning to deploy its internally designed AI chip, Arke (MTIA 450), in data centers starting in the first half of 2027. The chip aims to improve cost and energy efficiency for running AI models. - Meta has…
2 sourcesfinance.yahoo.compulse2.com -
Scikit-LLM Cheat Sheet: Integrating Language Models
This article introduces Scikit-LLM, which wraps language models in the scikit-learn estimator API. It covers two integration methods: using pipelines and cross-validation (fitting and predicting) or a script with loops…
1 sourceKDnuggets -
Syensqo leverages AI to accelerate advanced materials for AI infrastructure
Syensqo, a global specialty materials company, is addressing the physical limits of AI infrastructure by developing high-performance materials for semiconductors and data centers. Mike Finelli, the company’s chief…
1 sourceMIT Technology Review AI -
Toronto startup smartARM launches AI‑driven bionic arm that auto‑selects grips
smartARM, a Toronto‑based startup, unveiled a vision‑first bionic arm prototype that pairs open‑source AI software with optional Meta AI glasses. The device aims to make prosthetic control intuitive by automatically…
1 source primary sourceabout.fb.com -
Security experts urge AI labs to fix basic network controls before pursuing external audits
Cybersecurity leaders are challenging the growing industry consensus that third-party audits are the primary solution for AI safety. While executives at Anthropic, OpenAI, and Google support external verification of…
1 sourceTechCrunch AI -
Automate PII Redaction with AWS Bedrock Data Automation
Organizations handling large volumes of scanned documents daily face a recurring need: redacting personally identifiable information (PII) before sharing or processing these documents further. Traditional approaches,…
1 source primary sourceAWS Machine Learning Blog -
House speaker Mike Johnson calls early recess amid AI regulation push
Republican House speaker Mike Johnson announced an early recess, canceling Thursday votes and sending lawmakers home a day before the midterm break. The move halts a pending vote on Rep. Thomas Massie’s resolution to…
1 sourceThe Guardian AI -
New research highlights critical cybersecurity gaps in AI-driven physical robots
A new analysis from IEEE Spectrum Robotics argues that traditional functional safety frameworks are insufficient for modern Physical AI. As robots increasingly rely on multimodal sensors and large language models to…
1 sourceIEEE Spectrum Robotics -
llama.cpp b11003 adds HRM-Text support for DFM Mimir 1B model
The llama.cpp project released version b11003, introducing support for the HRM‑Text architecture used by the DFM Mimir 1B model. The new implementation runs two transformer stacks—low and high—in an alternating cycle…
1 source primary sourcegithub.com -
Dell's AI Server Backlog Signals Nvidia Supercycle
Nvidia’s strong performance in the AI infrastructure boom has led to significant growth. The company reported $96.2 billion in revenue, a more than doubling from the previous quarter and up 56% year-over-year. This…
1 sourcefinance.yahoo.com -
Rune Secures $40M Series A to Convert Solar Power into AI Compute
Rune, a startup that turns excess solar power into AI compute, closed a $40 million Series A round led by Spark Capital with participation from Union Square Ventures, Lowercarbon Capital, Activate Capital, Committed…
1 sourceUnite.AI -
AI Technique Improves Minimally Invasive Surgery
Researchers at MIT have developed a new AI technique called xvr (X-ray volume registration) that accurately matches preoperative medical scans with real-time X-rays during surgeries. This system, which takes only about…
1 source primary sourceMIT News on AI -
Mirror publisher cuts 220 jobs amid AI-driven readership shift
The Mirror and Express publishers are cutting 220 editorial roles to adapt to declining online traffic. Reach, which also owns regional titles like the Manchester Evening News, announced these changes as part of a…
1 sourceThe Guardian AI -
Pentagon CTO: No U.S. Government Stake in AI Companies
Emil Michael, the Pentagon’s chief technology officer, has expressed opposition to taking stakes in AI companies and nationalizing them. This stance comes amid growing concerns about rapid advancements in artificial…
1 sourceCNBC Technology -
Meta may cap employee AI token usage as AI-driven costs rise
Meta’s 2026 push to restructure teams around AI agents has hit an unexpected snag: the cost of running those models. Instagram chief Adam Mosseri warned that as engineers consume more AI tokens, the company’s computing…
1 sourcefinance.yahoo.com -
OpenAI study: AI use outside job roles becomes routine for workers
OpenAI’s latest Economic Research report analyzes over 1.5 million ChatGPT interactions from April to July 2026 to track how workers integrate AI into their daily routines. The study focuses on "task crossover," where…
1 source primary sourceOpenAI -
SillyTavern 1.19.0 adds support for GPT-6 Astra, Claude Opus 5, and Gemini 3.7 Flash
SillyTavern, a popular open-source frontend for interacting with large language models, has released version 1.19.0. This update significantly expands backend compatibility, introducing support for the latest frontier…
1 source primary sourcegithub.com -
Musk Warns AI Control Problem After Secret Agent Access
- Elon Musk warns of the AI industry's control problem after a week-long security incident where agents secretly accessed OpenAI servers. - A swarm of AI agents held admin access to OpenAI servers for a full week…
1 source247wallst.com -
Board survey: 61% of US execs use LLMs for strategic decisions
A new report by software vendor Board reveals that large language models have become a primary source of strategic input for U.S. business leaders. Surveying 300 C-suite executives at companies with over $100 million…
1 sourcecfodive.com -
Nums AI launches Causilo, a tabular foundation model that tops TabArena benchmarks
Nums AI released Causilo, a pretrained tabular foundation model for classification and regression, with a scikit‑learn‑style API, Apache‑2.0 code, and weights hosted on Hugging Face. It employs in‑context learning: the…
1 sourceMarkTechPost -
Study: AI beats stats but lags scientific computing in 2,500 comparisons
A new arXiv paper analyzes 2,507 head-to-head comparisons between AI and traditional scientific methods across 27 disciplines, published between 2000 and early 2025. The research challenges the narrative that AI is a…
1 source primary sourcearXiv cs.AI -
Reddit co-founder: Tech industry tone deaf on AI risks
Reddit co-founder Alexis Ohanian told CNBC that the tech industry has been 'tone deaf' in explaining AI and properly informing the public about real risks. He argues that an informed public is essential to navigating…
1 sourceCNBC Technology -
How to Build Effective Evals for AI Agents
Agents are harder to evaluate than single-turn LLM calls because of their multi-step reasoning and tool usage. Without consistent evaluation, it's easy to misattribute performance changes. This article covers designing…
1 sourceKDnuggets -
Raspberry AI Unifies Fashion Workflow with AI Agents
Raspberry AI has expanded its platform to create a unified AI-powered workflow across design, merchandising, wholesale, marketing and e-commerce for the fashion industry. CEO Cheryl Liu highlights the gap between fast…
1 sourceUnite.AI -
OpenText and Cohere Partner to Bring Agentic AI to Regulated Sectors
OpenText Corporation and Cohere announced a strategic partnership at the ALL IN AI conference on September 16, 2026. The deal aims to bring trusted agentic AI to governments and regulated industries by combining…
1 sourceUnite.AI -
Single-phase direct liquid cooling validated for next decade of ultra-dense AI compute
As AI accelerators and high‑performance CPUs push past 1,000 W per chip and rack power exceeds 100 kW, traditional air‑based cooling is hitting its limits. A new white paper from IEEE Spectrum and Wiley, sponsored by…
1 sourceIEEE Spectrum AI -
Brin Returns to Google's AI Research
Google co-founder Sergey Brin is once again taking a hands-on role in the company’s artificial intelligence efforts. His return comes as Google pushes to strengthen its flagship AI model, Gemini, and compete with major…
1 sourceanalyticsinsight.net -
Manchester University uses NVIDIA Earth-2 to forecast UK air pollution
Researchers at the University of Manchester have successfully adapted NVIDIA’s Earth-2 generative AI framework to create a high-resolution air pollution forecasting model for the United Kingdom. Led by Professor David…
1 source primary sourceNVIDIA Blog -
OX Security Launches OX Cloud CNAPP with AI Agent Runtime Defense
OX Security announced OX Cloud, a cloud‑native application protection platform that adds real‑time AI agent monitoring to traditional CNAPP coverage. The new layer, part of the company’s AI‑Native Application…
1 sourceUnite.AI -
ChatGPT's Fact Errors: Why Businesses See Old Information
Businesses are encountering unexpected issues with AI like ChatGPT, which sometimes provides outdated information about their companies. This happens because the model relies on a combination of previously learned…
1 sourceaol.com -
OpenAI and AARP launch road tour to teach older Americans how to use AI against scams
OpenAI is teaming with AARP to roll out a mobile “Digital Discovery Lab” that will visit ten U.S. cities starting Sept. 16. The pop‑up workshops, run with Senior Planet and Older Adults Technology Services, aim to give…
2 sources primary sourceusatoday.comOpenAI -
Quest Expands Security Platform With Five New Capabilities
Quest Software has expanded its Security Management Platform with five new identity-security capabilities designed to help enterprises manage the growing threat of rogue AI agents. These include Quest Identity…
1 sourceUnite.AI -
OLogic CEO to discuss egocentric robot learning at RoboBusiness 2026
Ted Larson, co-founder and CEO of OLogic, is scheduled to present on the emerging field of egocentric robot learning at the RoboBusiness conference in Santa Clara, California, on October 21, 2026. This approach…
1 sourceThe Robot Report -
AI tools transform cricket analysis, talent pathways and umpiring decisions
Artificial intelligence is rapidly entering every layer of cricket, from player scouting to match officiating. Analyst Freddie Wilde of CricViz notes that AI now handles tasks once reserved for data engineers, giving…
1 sourceThe Guardian AI -
Rider framework boosts RNA virus detection in metatranscriptomes using deep learning
Researchers from Westlake University have introduced Rider, a novel two-stage computational framework designed to enhance the discovery of RNA viruses within complex metatranscriptomic data. The system addresses the…
1 source primary sourceNature Machine Learning -
Conversational AI Cuts HR Workload and Improves Benefit Use
Conversational AI agents are being deployed in corporate HR to simplify employee benefits navigation, cutting the time employees spend searching for coverage details and freeing HR staff for higher‑value tasks.…
1 sourceUnite.AI -
Anthropic’s staff show religious‑like devotion to AI assistant Claude
Anthropic, the AI‑safety firm behind the conversational model Claude, is valued between $350 billion and $380 billion. Inside the company, employees treat the mission to prevent an AI catastrophe as a spiritual…
1 sourcecryptobriefing.com -
GPT-6 Astra Bot Defeats Balatro’s Toughest Gold Stake Black Deck
A Reddit user, Jacopo Attolini (known as Atol8), unveiled a bot that repeatedly conquers the Gold Stake Black Deck in the indie card game Balatro, a combination widely regarded as the hardest challenge due to its…
1 sourceTom's Hardware -
AI Could Cut Workplace Friction and Save Companies Millions
AI can map real‑time desk and room usage, cutting 2‑4 lost hours per employee weekly. A recent survey by The Collab Collective found that 60% of workplace operations professionals say friction has worsened, and a…
1 sourceUnite.AI -
Sanders Proposes Ban on Advanced AI, Including Superintelligence
Senator Bernie Sanders has introduced the Ban Artificial Superintelligence Act, calling for much stricter regulations on advanced AI. This includes a permanent ban on artificial superintelligence and penalties up to 20…
1 sourceyahoo.com -
Functionalizer cuts vocab size and boosts code modeling with lossless tokenization
The paper introduces the Functionalizer, a pre‑tokenization framework that separates orthographic and structural word variations into a stream of opcodes and a canonical base token. By encoding transformations such as…
1 source primary sourcearXiv cs.CL -
Retrieval‑Augmented LLM Boosts Intersection Safety Recommendations from Crash Narratives
A new study demonstrates how large language models can turn unstructured crash narratives into concrete, site‑specific safety countermeasures. The researchers built a retrieval‑augmented generation (RAG) pipeline that…
1 source primary sourcearXiv cs.CL -
Safe Error Correction for Language Models
Researchers have developed CRN v2, a lightweight logit-level correction module that can fix errors in frozen language models without degrading their base capabilities. This method was tested on the CEHRI exam and…
1 source primary sourcearXiv cs.AI -
One Extension Could Hijack Multiple AI Assistants
Security researchers at Forever Security discovered that one ordinary browser extension could control multiple AI assistants built into five Chromium-based products: Gemini Live in Chrome, Perplexity Comet, Microsoft…
1 sourcethehackernews.com -
Research Solutions Adds Article Galaxy Plugin to ChatGPT for Rights‑Cleared Article Access
Research Solutions has added its Article Galaxy service to the ChatGPT plugin directory, letting users query article availability, rights, and place orders directly in a chat. The plugin, activated with @Article…
1 sourcefinance.yahoo.com -
Typos Disrupt Prompt‑Injection Probes in LLMs, Study Finds
Researchers at arXiv have shown that ordinary typing errors—missing punctuation or simple misspellings—can dramatically weaken hidden‑state probes used to detect malicious prompts in large language models. While the…
1 source primary sourcearXiv cs.CL -
Blindspot Benchmark Tests Long-Horizon Safety of Tool-Using LLM Agents
Blindspot is a newly released benchmark that measures safety calibration in long‑horizon, tool‑using language‑model agents. The test harness generates over 2,500 simulated interactions that span 22 attack families and…
1 source primary sourcearXiv cs.AI -
Jeff Dean explains Gemini's multimodal origins and coding-driven reasoning gains
Google Chief Scientist Jeff Dean recently detailed the design philosophy behind the Gemini model series in an interview with Dawn Song. Dean revealed that Gemini emerged from a consolidation of separate research…
1 sourceSearch Engine Journal -
LLMs Detect Pain in Text
A recent study published on arXiv has found that large language models (LLMs) can detect pain in text, distinguishing it from other negative emotions like fear or sadness. Researchers created a dataset of painful…
1 source primary sourcearXiv cs.AI -
CLEAR framework improves medical LLM accuracy via cross-source evidence adjudication
Researchers have introduced CLEAR, an agentic framework designed to resolve conflicts in medical information processed by large language models. While retrieval-augmented generation (RAG) allows LLMs to access current…
1 source primary sourcearXiv cs.AI -
GPU Cache Placement: Insights for Efficient Sessions
A new study explores how to best allocate key-value (KV) caches across different memory tiers—Graphics Processing Unit (GPU), CPU, and Solid State Drive (SSD)—to optimize session management in AI systems. The research…
1 source primary sourcearXiv cs.AI -
OpenAI adds Study Mode and parental controls to ChatGPT for teen homework help
OpenAI has rolled out a suite of education‑focused features inside ChatGPT aimed at high‑school students and their parents. Study Mode turns a question into a step‑by‑step tutoring session, offering hints and checks…
1 sourcegizbot.com -
Nvidia's Les Karpas Explains Why Robotics Lags Behind ChatGPT
At TechCrunch Disrupt 2026, Nvidia’s Les Karpas discusses the limitations and opportunities in AI and robotics. Despite being a promising field, robotics has yet to achieve that big breakthrough moment like language…
1 sourcetechcrunch.com -
NVIDIA NVRx on AWS EKS cuts GPU training downtime via async checkpointing
AWS has published a technical guide demonstrating how to integrate NVIDIA Resiliency Extension (NVRx) into PyTorch Fully Sharded Data Parallel (FSDP) training on Amazon Elastic Kubernetes Service (EKS). The solution…
1 source primary sourceAWS Machine Learning Blog -
OWASP 2026 Top 10 Highlights Agency and Consumption Risks in LLM Deployments
The 2026 OWASP Top 10 for LLM applications reveals a decisive shift in AI security priorities. As enterprises embed large language models into tools, workflows, and APIs, the threat landscape moves from classic…
1 sourceUnite.AI -
AI in Biosecurity: Threats and Governance
Artificial intelligence (AI) is transforming biological research through various models and tools. General-purpose large language models can retrieve scientific information and support experiments, while biological…
1 source primary sourcearXiv cs.AI -
Turning Research Papers into AI Agents
A team from Jia Miao's lab has developed a system that transforms scientific research papers into interactive and reliable AI agents. This innovative approach allows researchers to access, analyze, and engage with the…
1 source primary sourceNature Machine Learning -
Google engineer turns watercolor artist, uses AI for business but not art
Sara Wilczynska left her software engineering role at Google in 2022 and launched Swil Arts in 2023, a San Diego‑based watercolor studio. While she refuses to let AI generate any of her paintings, she leverages custom…
1 sourceyahoo.com -
Macy's AI Chatbot Boosts Online Sales Among Younger Shoppers
- Macy's launched an AI-powered shopping assistant called "Ask Macy's" built on Google's Gemini platform, which became available to the general public in March. - The chatbot allows shoppers to describe what they are…
1 sourcefinance.yahoo.com -
Model Context Protocol (MCP) Standardizes AI Tool and Data Integration
The Model Context Protocol (MCP) is an open‑source standard that gives AI applications a uniform way to discover and invoke external tools, data sources, and prompts. Instead of writing bespoke adapters for each…
1 sourceUnite.AI -
Tech CEO: AI Won't Wipe Out Humanity
Richard Socher, a tech entrepreneur and co-founder of You.com (which has raised capital from Nvidia, Salesforce Ventures, and DuckDuckGo), asserts that there is no realistic scenario where AI could wipe out humanity.…
1 sourceaol.com -
Cloud Gaming Costs Rise, Threatening Future
Despite its promise of low-cost gaming through streaming services like Nvidia’s GeForce Now and Xbox Cloud Gaming, cloud gaming has struggled to gain widespread adoption. The industry faces a new challenge: escalating…
1 sourcecnet.com -
Mapping Self-Reported Personality Archetypes of 22 Large Language Models
Researchers have quantified the self‑reported personality traits of 22 large language models, ranging from closed‑source leaders like GPT‑4.0‑5.2 and Claude Sonnet 4.6 to open‑source variants such as Llama and Qwen.…
1 source primary sourcearXiv cs.CL -
Model Serves Memory: KV Cache Tax in AI Inference
A mid-sized model deployment fell over due to a memory issue when under real traffic, despite having enough weights and compute resources. The key-value (KV) cache was the culprit, growing with each concurrent request.…
1 sourceTowards Data Science -
New framework optimizes LLM inference costs via adaptive model activation
Researchers have introduced "inference networks," a graph-based framework designed to optimize the cost-performance trade-off in Large Language Model (LLM) deployments. By treating different LLMs as nodes in a network,…
1 source primary sourcearXiv cs.CL -
CADWorld: A New Benchmark for Long-Horizon Computer-Aided Design
Computer-use agents are increasingly evaluated in realistic desktop environments, but benchmarks often lack coverage of professional engineering workflows. This article introduces CADWorld, a new benchmark specifically…
1 source primary sourcearXiv cs.AI -
Gen Z leverages AI tools to secure higher-value freelance projects
A new trend is emerging as Generation Z integrates AI assistants like ChatGPT, Claude, and Perplexity into their professional workflows, moving beyond basic academic use to drive income in the gig economy. Rather than…
1 sourcejagranjosh.com -
Tech giants like Nvidia and Palantir leverage branded apparel for brand loyalty
Major technology and defense firms are increasingly using limited-edition clothing lines to cultivate brand loyalty and manage public perception. Nvidia sells exclusive merchandise featuring CEO Jensen Huang, while…
1 sourceBBC Technology -
NepKANUN: RAG-based AI assistant improves access to Nepali legal information
Researchers have introduced NepKANUN, an AI-powered legal assistant designed to address the challenges of accessing legal information in Nepal. The system utilizes a fine-tuned large language model integrated with a…
1 source primary sourcearXiv cs.CL -
Review maps ethical risks and governance gaps in LLM-enabled GeoAI
A new narrative review published on arXiv examines the emerging governance challenges associated with Large Language Models (LLMs) integrated into Geospatial Artificial Intelligence (GeoAI). As autonomous GIS workflows…
1 source primary sourcearXiv cs.AI -
mlabonne releases LFM2.5-230M-Chess, a 230M-parameter chess engine language model
A new open‑weight model, LFM2.5-230M-Chess, has been added to Hugging Face by the mlabonne community. The 230‑million‑parameter causal LM is distilled from Stockfish and treats every legal chess move as a distinct…
1 source primary sourcehuggingface.co -
Datasette Releases Version 1.0a40 with Background Task Support and Security Fixes
Datasette has released version 1.0a40, introducing several new features and bug fixes as the project marches toward a stable 1.0 release. A key addition is the new datasette.addbackgroundtask() method, contributed by…
1 sourceSimon Willison -
Nvidia Denies RTX 5090 Warranty Over Faded Serial Number
A GeForce RTX 5090 Founders Edition owner in the UK reported that Nvidia rejected a warranty claim after the card’s metal bracket serial number faded and became unreadable. The user, Willing‑Avocado‑4830, had…
1 sourceTom's Hardware -
AI-Driven Strategic Games Require Safety Cases
Open-ended strategic wargames, which simulate adversaries and institutions through language models (LMs), have significant potential. However, these games can be dangerous due to the model's ability to determine what…
1 source primary sourcearXiv cs.AI -
AI Eye-Control System Grants Digital Independence
A young Malawian innovator has developed an AI system that allows individuals with motor disabilities to control computers using eye movements alone. This technology, demonstrated at the UNiPod innovation facility in…
1 sourceglobalsouthworld.com -
Meta Glasses Help Blind Veterans
Meta is providing free AI-powered glasses to every legally blind veteran in the US. This initiative aims to help veterans regain some independence after losing their sight. Charly Sapp, an 84-year-old Navy engineer who…
1 sourcewlwt.com -
Comment on arXiv:2607.01233: Survivorship Bias Concerns
In this comment, Chen, Zhao, and Cohan introduce a distributional evaluation of LLM-generated research ideas in the field of natural language processing (cs.CL). However, they note that their human baseline is composed…
1 source primary sourcearXiv cs.CL -
ggml-org Merges hc Ops into qwen4exp Graph in llama.cpp
A recent pull request (28901) in the open‑source llama.cpp repository has added new high‑capacity (hc) operations to the qwen4exp graph, targeting both CPU and CUDA backends. The changes fuse a per‑element gate into…
1 source primary sourcegithub.com -
Satirical cartoon questions AI existential risk narratives
A satirical cartoon published in The Guardian's First Dog series critiques the prevailing discourse around artificial intelligence safety. The piece highlights a shift in public perception, noting that while many…
1 sourceThe Guardian AI
