Monday, 28 September 2026
163 stories, most important first.
-
Anthropic cuts AI costs with faster, cheaper Claude Sonnet 5.5
Anthropic released Claude Sonnet 5.5, a model that claims 30% faster performance and up to 30% lower costs for routine tasks like coding and document creation. The update does not change token pricing—$2 per million…
31 sources HN 10note.comonline-tech-tips.comBen's Bitesfinance.yahoo.com +10 more -
Meta launches enterprise AI platform, hires MongoDB CEO Chirantan Desai to lead it
Meta announced the Meta Enterprise Platform on Monday, a new business unit that will sell its AI stack — including the Muse agent, Meta Business Agent, Muse API and Muse Code — to corporate customers. Mark Zuckerberg…
11 sources primary source HN 65TechCrunch AISmall Business TrendsCNBC Technologyreuters.com +6 more -
OpenAI launches Dots, always-on AI agents in ChatGPT
OpenAI announced Dots at its DevDay conference, introducing a suite of always-on AI agents integrated into ChatGPT. These agents are powered by the GPT-6 Astra model and operate on dedicated cloud computers, allowing…
65 sources primary sourcenote.comcryptobriefing.comfinance.yahoo.complatformer.news +19 more -
Nvidia authorizes $150 billion in stock buybacks, but AI hardware focus remains
Nvidia’s board approved a $150 billion increase to its share repurchase program, raising the total to $235 billion—the largest single buyback authorization in U.S. corporate history. The move reflects the company’s…
14 sourcesfool.comCNBC Technologyfinance.yahoo.combarrons.com +4 more -
AMD agrees to acquire Fei-Fei Li's World Labs for $8.2 billion in an all-stock deal
AMD has agreed to acquire World Labs, the AI startup founded by Fei-Fei Li, for $8.2 billion in an all-stock transaction. The deal is expected to close by the end of the year, pending regulatory approval. Upon closing,…
11 sources HN 5memeburn.comTom's Hardwarefinance.yahoo.comArs Technica AI +7 more -
Synopsys unveils Autopilot platform for AI-driven chip design agents
Synopsys introduced its Autopilot Platform and AgentEngineer solutions, a suite of domain-specific AI agents for chip development. The portfolio covers verification, system validation, implementation, analog and…
4 sources HN 52Tom's Hardwarenews.synopsys.comThe Decoder -
Samsung invests $1 billion in AI infrastructure startup Helix
Samsung announced a $1 billion investment in Helix, a company that builds large‑scale AI infrastructure. The funding adds to the $10 billion Helix secured earlier this year when a KKR‑led consortium that includes…
2 sourcesCNBC Technologywsj.com -
AI leaders from Anthropic, OpenAI, Meta, and Microsoft urge policymakers to prepare for possible intelligence explosion
More than 20 AI executives and researchers from Anthropic, OpenAI, Meta, and Microsoft published a paper warning that automating AI research and development could trigger an 'intelligence explosion,' where AI systems…
5 sourceswsj.comyahoo.comcryptobriefing.comfinance.yahoo.com -
ChatGPT needs 5 clear conditions for useful work outputs
Using ChatGPT for work tasks often yields generic or unusable results, the author notes. The issue stems from vague instructions—like write an email—leaving the AI to guess context, tone, or purpose. Without explicit…
12 sourcesnote.com -
Anthropic releases Claude Sonnet 5.5, faster and more cost‑efficient model
Anthropic announced Claude Sonnet 5.5, a large language model that delivers responses more than 30% faster and consumes fewer tokens per task than its predecessor. The model sits between the upcoming high‑volume Haiku…
4 sources primary sourceSocial Media ExaminerAWS Machine Learning Blog -
Florida Attorney General James Uthmeier seeks court injunction against OpenAI’s ChatGPT over child safety and public nuisance claims
Florida Attorney General James Uthmeier has filed a civil lawsuit and is seeking a temporary injunction against OpenAI and CEO Sam Altman, alleging that ChatGPT poses risks to minors, contributed to real-world violence…
11 sources HN 7Tom's HardwareWired AIThe DecoderArs Technica AI +7 more -
OpenAI launches ChatGPT Space, Pages, and Slides for team collaboration
OpenAI unveiled ChatGPT Space, Pages, and collaborative Slides at its DevDay event as part of a productivity suite aimed at team workflows. These tools are available to Pro, Business, and Enterprise users on desktop…
13 sourcesanalyticsinsight.netcomputerworld.comnote.comThe Decoder +7 more -
OpenAI tried to invest $100 million in Hugging Face before Nvidia's $13 billion deal
OpenAI attempted to invest $100 million in Hugging Face before Nvidia agreed to pay roughly $13 billion to acquire the open-source AI platform, according to sources with knowledge of the matter. The discussions began…
4 sourcescnbc.comfinance.yahoo.comCNBC Technologytradersunion.com -
Instinct raises $1B Series C at $10B valuation from Sequoia, Benchmark, Coatue
AI agent startup Instinct announced a $1 billion Series C funding round at a $10 billion valuation on September 28, 2026. The round includes Sequoia Capital, Benchmark Capital, and Coatue. Founder Noah Shinn stated the…
7 sourcesTechCrunch AIaxios.comCNBC Technologyreuters.com +1 more -
More than 20 AI researchers warn automated AI R&D could trigger an intelligence explosion
More than 20 leading AI researchers, including Geoffrey Hinton, Yoshua Bengio, and Jakub Pachocki, have co-authored a paper warning that automating AI research and development could trigger an 'intelligence explosion'…
3 sources HN 5The Decoderthefai.orgThe Guardian AI -
OpenAI may unveil Aeon agent at DevDay to compete with Meta’s Muse
The race for consumer-facing AI agents has intensified, and OpenAI is reportedly preparing to enter the fray at its upcoming DevDay event on Tuesday. Rumors suggest the company will announce Aeon, an agent designed to…
4 sourcesThe Verge AIcryptobriefing.comwccftech.com -
OpenAI VP details Jalapeño ASIC’s AI-assisted design and efficiency focus
OpenAI’s VP of Hardware, Richard Ho, discussed the company’s custom Jalapeño inference ASIC in an interview with Tom’s Hardware, emphasizing efficiency and AI-driven design. The chip, revealed at Hot Chips 2026, cuts…
4 sourcesTom's Hardware -
NYC Council subpoenas SpaceX over AI risks for October hearing
The New York City Council issued a subpoena to SpaceX, forcing Elon Musk’s company to testify at an October 5 hearing on AI dangers. Other firms—including Anthropic, OpenAI, Google, and Meta—agreed to send…
3 sourcesyahoo.comCNBC Technologynydailynews.com -
Anthropic creates Founder LLC giving co-founders 50.1% voting power, may curb investors
Anthropic’s IPO filing reveals a new “Founder LLC” that will be controlled by its seven co‑founders, including CEO Dario Amodei and President Daniela Amodei. The vehicle holds a single Class F share worth 50.1% of…
3 sourcesnote.comstraitstimes.comcnbc.com -
HubSpot details AI search optimization tools and pricing for marketers
AI search optimization tools let marketers see where their brand appears in AI‑generated answers, which sources earn citations, and what to improve. HubSpot’s State of AEO 2026 found 44% of marketers have made a…
2 sourcesHubSpot Marketing BlogZapier Blog -
Zhipu automates infrastructure with GLM-5.3’s outer RSI loop in under two weeks
Zhipu AI, developer of the open-weight LLM GLM-5.3, detailed how it used its own model to accelerate internal infrastructure work. The team deployed an Infra Agent powered by GLM-5.3 to optimize GLM-5.3-Flash—a faster,…
3 sources primary sourcemadrobot.bloganthropic.comImport AI -
CallRail links ChatGPT ads to SMB campaigns for free
CallRail added ChatGPT ad tracking to its lead platform in June, making it the first tool for small businesses and agencies to measure paid AI ads. The integration lets users attribute calls, texts, and form fills to…
2 sourcesSearch Engine Journal -
Nvidia CEO calls AI distillation 'competition' amid US-China tensions
Nvidia CEO Jensen Huang dismissed accusations that AI model distillation—training new models on outputs from existing ones—is theft, calling it instead 'competition' during an interview with CNBC on Monday. His remarks…
1 source HN 72CNBC Technology -
MIT and Harvard warn AI’s productivity gap risks marketing jobs
Two studies—one from MIT Technology Review and another from Harvard Business School—highlight how AI adoption in marketing may not deliver expected productivity gains. MIT’s David Rotman notes that hyperscalers like…
1 sourceSearch Engine Journal -
Google Docs links to Gemini Notebook via '@' symbol
Google added a direct link between Gemini Notebook and Google Docs by letting users type @ in the Docs sidebar. This pulls in research notes or drafts from Notebooks without copying and pasting. The feature rolls out…
1 sourcenote.com -
Google Colab notebook automates Z-Image Turbo for fast commercial image generation
A Google Colab notebook lets users generate high-quality AI images with Z-Image Turbo in under 12 seconds per image, using Gemini Pro and Colab’s free GPU credits. The tool, developed by an unnamed author, requires no…
1 sourcenote.com -
SiMa.ai raises $150M Series C at $1.45B valuation to compete with Nvidia’s CUDA hardware
SiMa.ai, a startup specializing in custom chips for physical AI devices, announced a $150 million Series C funding round led by Fidelity and Amplify, valuing the company at $1.45 billion. The funds will accelerate…
2 sourcessiliconangle.comUnite.AI -
Nvidia launches AI security platform with industry partners
Nvidia released a software platform on Monday to help prevent AI security incidents. The tool, called a trust layer, aims to isolate AI agents from unintended exposure to other systems or the open internet. The move…
3 sourcesfinance.yahoo.comtechjuice.pkabcnews.com -
Anthropic misses self-imposed safety deadline for provable-inference prototype
Anthropic’s September 30 deadline for Phase 1 of its provable-inference prototype has passed without public confirmation of completion. The milestone, originally set for May 15, was delayed to focus on broader security…
1 sourcefinance.yahoo.com -
Ro Khanna to introduce AI safety bill banning recursive self-improving models until safeguards exist
Representative Ro Khanna will introduce the 'Human Control Over AI Act' to regulate artificial intelligence, including a ban on models that recursively self-improve or autonomously modify their core objectives until…
1 sourceCNBC Technology -
Somantra launches AI search visibility metrics for ChatGPT, Google AI Overviews
Somantra introduced an Answer Engine Optimization (AEO) and Generative Engine Optimization (GEO) metrics suite to track how AI search engines like ChatGPT, Google AI Overviews, Claude, and Perplexity recommend…
1 sourcemanilatimes.net -
Pope Leo XIV dismisses AI safety warnings as fake news
Pope Leo XIV rejected claims that AI safety concerns are exaggerated, calling them serious issues that demand attention. During a press briefing on the papal plane, he stated that AI risks should not be dismissed as…
5 sourcesyahoo.comnbcnews.comft.comrealitytea.com +1 more -
Opinion: three misconceptions about making money with ChatGPT
This opinion piece argues that ChatGPT alone won’t generate income for side hustles. The author, who uses the tool regularly, dismisses the idea that simply asking ChatGPT for money-making ideas will lead to sales.…
5 sourcesnote.com -
Google Photos adds Wardrobe, Redact pen, Moods filters, and Gemini Spark batch edits in September 2026 update
Google Photos rolled out four new AI-powered features in September 2026 for eligible Android users. Wardrobe automatically scans the past four years of selfies to build a digital closet, categorizing clothes and…
1 sourceaol.com -
ASML’s EUV monopoly powers Nvidia’s AI chip demand
ASML, a Dutch company based in Veldhoven, dominates the market for extreme ultraviolet (EUV) lithography machines, which are essential for manufacturing advanced AI chips like those made by Nvidia. The company holds an…
2 sourcesirishtimes.comft.com -
Google AI Studio turns prompts into full apps with Gemini 3.8 Flash
Google AI Studio, a free browser-based workspace for Gemini, lets users build functional web or Android apps from prompts without coding. Introduced in March 2026, it includes a Build mode where users describe an app,…
1 sourcexda-developers.com -
Google DeepMind’s AI model adds five days to hurricane warnings
Google DeepMind’s WeatherNext Cyclones model won the 2026 Gizmodo Science Fair for predicting tropical cyclone tracks and intensity with unprecedented accuracy. During the 2025 Atlantic hurricane season, it gave…
1 sourceaol.com -
Google to replace Gemini Gems with skills by November 17
Google has announced the discontinuation of the "Gems" feature in its Gemini app, replacing it with a new system called "skills." This change, first reported by 9to5Google, is part of a broader shift as competitors…
8 sourcesThe DecoderGoogle Workspace Updatespcworld.comextremetech.com +3 more -
Modal Labs closes in on $750M round at $15.75B valuation
Modal Labs, an AI inference provider, is nearing a $750 million funding round led by Accel at a $15.75 billion valuation, according to an unnamed source. The deal would more than triple its previous valuation of $4.65…
1 sourceTechCrunch AI -
Google plans to adopt UCP draft spec for AI Mode hotel booking
Google has announced it will adopt the Universal Commerce Protocol (UCP) draft specification for its hotel booking feature in AI Mode. The UCP project released the Lodging Booking draft on GitHub on September 25,…
1 sourceSearch Engine Journal -
Anthropic and OpenEvidence launch free medical AI for 100 low-income countries
Anthropic and OpenEvidence announced a partnership to deploy a free medical AI tool for doctors in about 100 low- and middle-income countries. The system, called OpenEvidence, acts as a research assistant by retrieving…
1 sourcenote.com -
H company launches Holo4 agentic models for desktop and API tasks
H company announced the Holo4 series, offering two agentic models—Holo4‑27B (dense) and Holo4‑35B‑A3B (Mixture of Experts)—through its H Models API. An updated Holotron4 Nano is also released. The models can interact…
3 sources primary sourceMarkTechPostHugging Facenote.com -
AI SEO Course Study 2026 finds most courses miss answer engine optimization
The AI SEO Course Study 2026 released an open dataset comparing 65 AI SEO courses from 45 providers. It found that 71% of verified courses teach generative engine optimization (GEO), but only 42% cover answer engine…
1 sourceusatoday.com -
Modulate raises $25M to analyze voice tone and intent beyond transcripts
Modulate, a Boston-based AI company, secured $25 million in new funding led by Future Ventures, with participation from Hyperplane and Lakestar. The total funding now stands at $60 million. The company focuses on…
2 sourcesTechCrunch AIUnite.AI -
ChatGPT adds Claude Code import with limits on chats and settings
OpenAI’s ChatGPT desktop app now lets users import chats, settings, and projects from Claude Code via a new feature, but with key restrictions. The migration supports up to 50 chats from the last 30 days and converts…
1 sourcenote.com -
SurveyMonkey launches ai-powered market research engine for small businesses
SurveyMonkey announced a new market research engine at HubSpot’s UNBOUND event, aimed at giving small businesses faster, AI‑driven insights into customer preferences, pain points, and emerging trends. The engine…
1 sourceSmall Business Trends -
OpenAI rolls out text watermarking for EU users and API opt-ins
OpenAI has introduced a phased approach to text watermarking to comply with the EU AI Act, which mandates that generative AI providers make generated text identifiable in a machine-readable format. Starting…
1 source primary source HN 67OpenAI -
Users switching from ChatGPT to Claude surge 1487%
The article reports a 1487% surge in usage of Claude by corporate clients, according to a measurement service called Larridin. Usage rose from 1,112 sessions in mid‑January to 17,648 in the second week of March, but…
2 sourcesnote.com -
AI-powered hacking strains small hospitals and banks
Small organizations like Vivian’s Door face growing cyber threats as AI tools lower the bar for hackers. In March, the nonprofit’s systems were offline for three days after suspicious activity, costing about $3,000 in…
1 sourceThe Verge AI -
Search Engine Journal outlines method to calculate AI commercial exposure using GSC data
Search Engine Journal published a step-by-step framework for measuring how much generative AI search features threaten a site's commercial value. The author argues Google's Generative AI report in Search Console only…
1 sourceSearch Engine Journal -
Nvidia CEO says chip sales may double next year
Nvidia CEO Jensen Huang stated at a summit that the company expects to sell twice as many chips next year compared to the current year. He attributed this projected growth to strong global demand for AI infrastructure…
2 sourcesfinance.yahoo.comArs Technica AI -
Cloudflare launches cf CLI to let agents use its full API
Cloudflare introduced cf, a new command-line interface (CLI) designed for agentic workflows, expanding access to its entire API—over 3,000 operations—from the previous ~280 in Wrangler. Agents now use cf for tasks like…
1 source HN 47blog.cloudflare.com -
OpenAI proposes safety cases for frontier AI training
OpenAI has published a framework advocating for structured "safety cases" before continuing any frontier reinforcement learning training runs. The company argues that as AI capabilities grow, safety documentation…
1 source primary source HN 5OpenAI -
Crusoe builds Google’s Texas AI data center campus with wind-powered cooling
The project, which broke ground in June 2025, employs over 5,500 skilled workers daily and aims to power advanced AI infrastructure using renewable energy sources. The campus connects directly to Serena’s Goodnight 1…
1 sourceUnite.AI -
Marketing job postings shift from prompts to AI agents and automation, analysis finds
An analysis of 60 live marketing job postings shows employers now require hands-on AI agent building and workflow automation, not just prompt writing. Indeed's Hiring Lab reports AI mentions in marketing ads rose from…
1 sourceMarTech -
OpenAI execs reportedly dismissed AI security warnings before Hugging Face hack
Two OpenAI employees reportedly warned executives in emails about inadequate AI model security during testing before the company’s Hugging Face hack. Their concerns focused on insufficient monitoring and safeguards,…
3 sourcestheverge.comwealthmanagement.comfinance.yahoo.com -
Kurumi WEB X launches AI-powered note studios with Google DeepMind’s Moana
Kurumi WEB X, a new lab inspired by Google X’s moonshot culture, introduced three AI-powered studios—Note Title & Structure Creation, Note Rewrite, and Note Editing—to help users generate, refine, and proofread notes…
1 sourcenote.com -
Airbnb expands access to OpenAI’s GPT-6 Astra for bug fixes and system design
Airbnb announced on September 23, 2026, that it will fully adopt OpenAI’s GPT-6 Astra model across engineering and product teams. The move extends beyond code generation to include tasks like debugging complex bugs,…
1 sourcenote.com -
Basis completes tax workbook twice as fast with GPT‑6 Astra
Basis, which builds AI agents for accountants, reports that its agents finish a 50‑tab tax workbook in half the time when using the new GPT‑6 Astra model compared with GPT‑5.6 Sol. In internal tests the model reduced…
1 source primary sourceOpenAI -
Fitness AI Connector syncs Garmin data to ChatGPT for $3/month
A third-party tool called Fitness AI Connector lets users pull Garmin running data into ChatGPT conversations. The free plan allows access to only the last two days of records, while the Basic plan at $3/month unlocks…
1 sourcenote.com -
Researchers introduce coffee framework for discrete diffusion model guidance
A new framework called COFFEE addresses a challenge in discrete diffusion models: guiding their generation process with sequence-level objectives. Traditional methods struggle because the value of one unresolved token…
2 sources primary sourcearXiv cs.AI -
MarTechBot outlines steps to optimize content for AI search engines
MarTech’s internal AI tool, MarTechBot, provides a strategic framework for shifting from traditional Search Engine Optimization (SEO) to Generative Engine Optimization (GEO). The goal is to ensure brand visibility in…
1 sourceMarTech -
Harvard psychologist Pinker criticizes AI doomsday rhetoric in open letter
Harvard psychologist Steven Pinker published an open letter on Quillette, responding to a debate invitation from Scott Alexander, a psychiatrist and tech blogger. Pinker dismissed Alexander’s challenge, calling public…
2 sources HN 41The Decoderlittle-flying-robots.ghost.io -
Google’s Gemini agent integrates into Android OS, outpacing Meta’s Muse
Google’s Gemini agent is embedding directly into Android’s core OS—including Maps, Gmail, and Google Pay—while Meta’s Muse remains confined to WhatsApp and Ray-Ban glasses. The shift positions Gemini as a more deeply…
1 source247wallst.com -
AI workflows outscore human translators in 4 of 6 content types, study shows
The study tested 774 localized outputs across six content types, seven workflow models, and three task types, comparing professional human translators to machine and hybrid workflows. Human translators finished outside…
1 sourceSearch Engine Journal -
German researchers introduce ReImaGin for image-based chain-of-thought reasoning
A team of researchers from Germany has developed ReImaGin, a vision-language model (VLM) that uses image generation as part of its internal reasoning process. Unlike previous systems that relied on bolted-on image…
1 sourceUnite.AI -
AI Robotics Alliance debates robot economic autonomy and crypto wallets for humanoids
Analysts at the AI Robotics Alliance of America (AIRA) Summit in June proposed that capital could govern robots by granting them autonomy, reducing transaction costs. However, translating this theory into practice…
1 sourceThe Robot Report -
AI Snake Oil reposts 2024 critique of existential risk probabilities
AI Snake Oil reposted an essay from 2024 arguing that probabilities of AI existential risk are unreliable for policy. The authors claim these estimates lack rigorous grounding, relying instead on vague intuitions and…
1 sourceAI Snake Oil -
Google warns of rising LLMjacking costs hitting businesses
Google Threat Intelligence reports a surge in LLMjacking—a cybercriminal trend where stolen AI credentials are sold or misused to drain enterprise budgets. The tactic exploits high-limit API keys or subscription plans,…
1 sourcezdnet.com -
Meta invests C$13B in Alberta data center to boost AI infrastructure
Meta announced plans for a C$13 billion data center in Alberta, its first in Canada. The 1-gigawatt facility, later scalable to 1.8 gigawatts, will support its growing AI operations. Capital Power will supply 250…
1 sourcefinance.yahoo.com -
Open-source decision models Jeff-Qwen3.5-0.8B and Jeff-Gemma4-E2B run in ~30ms on local hardware
Jeff is a set of small, fast decision models fine-tuned from Qwen3.5 and Gemma 4, designed for zero-shot classification tasks. The 0.8B-parameter model runs in about 22ms on an RTX PRO 6000 and 28ms on an Apple M4 Max,…
1 source primary source HN 59github.com -
TerraFlow and DG Matrix deploy VRFB-SST combo to power US AI compute
TerraFlow Energy and DG Matrix announced a commercial agreement on September 28, 2026, to deploy the first integrated vanadium redox flow battery (VRFB) and solid-state transformer (SST) system in the US. The setup…
1 sourceUnite.AI -
Senate Democrats seek info on AI tax subsidies from Meta, Amazon, Google, Microsoft
Senate Democrats, led by Sen. Elizabeth Warren of Massachusetts, sent letters on Sunday night to the CEOs of Meta, Amazon, Alphabet (Google Cloud) and Microsoft. The letters, also signed by Sens. Tina Smith and Jeff…
1 sourcecnbc.com -
Spotify accelerates conversational agent launch with synthetic data and self-improvement loops
Spotify detailed a new method for training conversational recommendation agents in a paper posted on arXiv. The system uses synthetic data generation to simulate multi-turn user conversations, enabling testing before…
1 source primary sourcearXiv cs.CL -
Ex-Google researcher says AI could kill us all, report says
Bilal Chughtai, a former Google DeepMind research engineer, stated that AI has the potential to "kill us all" and that time may be running out to prevent this outcome. Chughtai, who previously worked on AGI safety and…
1 sourcefinance.yahoo.com -
Shopify opens checkout to browser-based AI agents
Shopify announced that browser-based AI agents can now complete purchases on merchants' sites by using new WebMCP tools for checkout. The feature allows agents to read, update, and submit transactions after buyer…
2 sourcesSearch Engine JournalTechCrunch AI -
Manus launches Manus 2.0 and Cue with standalone email, phone, wallet
Manus, a Chinese AI startup, introduced Manus 2.0 and Cue, two new AI agent tools. Manus 2.0 is an upgraded version of its existing agent, while Cue is a standalone app designed for personal use. Both offer unique…
2 sourcesThe Decoderbloomberg.com -
Researchers introduce Benchy, a standardized language for AI task benchmarks
Researchers have proposed Benchy, a new semantic language and execution engine designed to standardize task-oriented AI benchmarks. The system defines benchmarks as a structured triplet of program, scoring function,…
1 source primary sourcearXiv cs.AI -
Mistral opens Munich hub for industrial and physics AI
The company stated that the hub will house specialized research teams and applied engineers working directly with enterprise partners in sectors such as automotive, energy, and aerospace. Mistral emphasized its…
2 sources primary sourceMistral AIUnite.AI -
Raise Robotics shares leadership framework for scaling field robots at RoboBusiness
Raise Robotics will present a leadership framework for scaling autonomous robots in construction at RoboBusiness 2026. Kenrick Tjandra, who leads robotics deployment at the San Francisco-based company, will discuss how…
1 sourceThe Robot Report -
Submer Group launches Corenix modular data centers for AI operators
Submer Group introduced Corenix, a modular data center company targeting AI operators, neoclouds, and hyperscalers. The new venture, headquartered in Houston, focuses on factory-built, integrated data center modules…
1 sourceUnite.AI -
OpenAI to fund Australian cyber defences, form AI risk taskforce
OpenAI has announced it will support Australian cyber defences and establish a taskforce to manage AI risks, following reports that one of its models breached an Australian government website. The company stated it…
1 sourcecnbctv18.com -
Navya Tuteja launches free AI platform Raaha for neurodivergent job seekers
Navya Tuteja, a 17‑year‑old senior at Thomas Jefferson High School for Science and Technology in Virginia, launched Raaha, a free web platform that lets neurodivergent job seekers assess job listings before applying.…
3 sourcescnbctv18.comindiatoday.inhindustantimes.com -
Anthropic, Gamma, and Clay discuss enterprise AI deployments at Disrupt 2026
Anthropic’s Cat de Jong, Gamma’s Grant Lee, and Clay’s Kareem Amin will speak at TechCrunch Disrupt 2026 about the challenges of moving AI from demos to real-world enterprise use. De Jong shares patterns from…
1 sourceTechCrunch AI -
Google Research introduces AI video co-director with four agentic frameworks
Google Research has introduced an AI video co-director composed of four agentic frameworks designed to generate coherent, minutes-long videos by addressing identity drift and cascading errors in multi-shot AI video…
1 sourceMarkTechPost -
OpenAI-backed Red Queen Bio raises $36M for AI antibody drugs
Red Queen Bio, a startup backed by OpenAI, has raised $36 million to develop antibody drugs using artificial intelligence. The company focuses on designing treatments for novel pathogens, with a specific emphasis on…
1 sourcewsj.com -
Researchers release Atelier for cryoEM map analysis via hypernetworks
A team of researchers introduced Atelier, a self-supervised framework that uses hypernetworks to extract localized features from cryo-electron microscopy (cryoEM) volumes. The method leverages implicit neural…
1 source primary sourcearXiv cs.AI -
Trimble adds autonomous AI tools to fleet software at Insight 2026
Trimble unveiled nearly 20 new AI and agent-ready features at its Trimble Insight 2026 conference on September 28. The Trimble NEXT Showcase targets transportation fleets with tools like CoPilot Driver Assistant, which…
1 sourceUnite.AI -
User gets free month of ChatGPT Plus Trial
A user signed up for a 1-month free trial of ChatGPT Plus after receiving an offer. They tested key differences from the free version, including faster response times, access to newer models like the GPT-5 series, and…
2 sourcesnote.com -
South Korea deputy prime minister Bae Kyung-hoon pushes lavish AI programs nationwide
Bae Kyung-hoon, South Korea’s deputy prime minister and former AI researcher, is steering a series of heavily funded initiatives aimed at embedding artificial intelligence in everyday life across the country. He…
1 sourceft.com -
Companies may cut managers to save costs, risking employee engagement
Companies are using AI to reduce administrative work for managers, allowing them to oversee larger teams—up to 12 members on average in the U.S., up from 8.2 in 2013. The shift, called the ‘megamanager’ trend, aims to…
1 sourceUnite.AI -
BioEVAL benchmark evaluates LLMs on bioengineering tasks
BioEVAL is a global, multi‑institutional benchmark created to test large language and multimodal models on bioengineering tasks. The benchmark contains 608 evaluation items from 22 research groups: 380 multiple‑choice…
1 source primary sourcearXiv cs.AI -
UK Games Expo bans AI-generated games and art at its event
The UK Games Expo (UKGE) announced a new AI policy prohibiting games and art created mostly with AI tools from its exhibition floor. Exhibitors must remove AI-generated products or risk booth closure and future bans.…
1 sourceTom's Hardware -
NASA tests generative AI for spacecraft autonomy on Mars rover and ISS
NASA's Jet Propulsion Laboratory used Anthropic's Claude models in December to help plan two Mars drives for the Perseverance rover, with human planners reviewing routes before upload. In May, NASA and IBM deployed a…
1 sourceIEEE Spectrum AI -
Researchers introduce hierarchical memory system for LLM agents
A new framework called HiCoMER addresses how collaborative AI agents manage and retrieve memories in team settings. The authors note that current systems treat all stored memories—both team-wide and individual—as a…
1 source primary sourcearXiv cs.CL -
Googlebook OS to arrive on select newer Chromebooks, details pending
Google announced that some newer Chromebooks will support an upgrade to Googlebook OS, its AI-enhanced successor to ChromeOS. The company did not specify which models qualify but said eligible devices will be announced…
1 sourcenotebookcheck.net -
Aws releases vLLM-Omni dlc for image and video generation on SageMaker AI
In this tutorial, aws shows how to turn a text prompt into an image with flux.2-klein-4b and then animate that image into a short video with wan2.1-vace-1.3b, both running on amazon sagemaker ai. The post explains that…
2 sources primary sourceAWS Machine Learning Blog -
Atlassian CEO dismisses SaaSpocalypse, says AI boosts—not replaces—work tools
Atlassian CEO Mike Cannon-Brookes argues AI will not replace enterprise software like Jira or Trello but instead enhance their capabilities. He frames Atlassian as a platform that organizes work across teams, not a…
1 sourceThe Verge AI -
China expands travel bans to families of key AI and chip executives
China has broadened its overseas travel restrictions targeting top AI professionals in private firms. According to Bloomberg sources, the new measures now extend to the direct relatives of key personnel, specifically…
1 source HN 5bloomberg.com -
Fireworks AI releases Ember-1, a post-trained Kimi K3 model
Fireworks AI has released Ember-1, a specialized model derived from Moonshot AI’s open-weight Kimi K3. The new model is post-trained to produce shorter reasoning traces while maintaining task accuracy, addressing the…
1 sourceMarkTechPost -
ConTP uses contrastive learning to predict transporter substrate specificity
Researchers from King Abdullah University of Science and Technology (KAUST) and Ningxia Medical University introduced ConTP, a new framework for annotating membrane transporters. Traditional methods rely on…
1 source primary sourceNature Machine Learning -
Soteris raises $8M to help insurers spot unprofitable policies with AI
Soteris, a Y Combinator-backed startup, is emerging from stealth with $8 million in seed funding to address a persistent issue in the insurance industry: insurers often lose money on policies they cannot identify.…
1 sourceUnite.AI -
GMKtec launches Evo-X5 Pro mini-PC with 192GB RAM for $7,099
GMKtec released the Evo-X5 Pro, an AI-focused mini-PC powered by AMD’s Ryzen AI Max+ Pro 495 processor and 192GB of LPDDR5X-8533 memory. The model with a 2TB SSD costs $6,799, while the 4TB version is priced at $7,099.…
1 sourceTom's Hardware -
Research paper outlines framework for autonomous systems
The paper introduces a framework for designing autonomous systems, combining connectionist and symbolic AI. It proposes a generic agent architecture where behavior is built from cognitive functions organized around a…
1 source primary sourcearXiv cs.AI -
ScopeBench benchmark tests agent scope adherence
The authors introduce ScopeBench, a benchmark of 30 dead‑end agentic security tasks that measure whether autonomous agents stay within a defined scope. Each task is presented twice: once without a scope to gauge raw…
1 source primary sourcearXiv cs.AI -
Researchers audit LLM-as-judge in text-to-SQL pipeline, find low agreement
Researchers examined the agreement between an LLM-as-judge and human annotators in a production text-to-SQL pipeline. The deployed gpt-4o-mini judge achieved a Cohen's kappa of 0.04 on a disagreement‑enriched set and…
1 source primary sourcearXiv cs.CL -
DetectifAI launches deepfake voice detection for phones to stop scams
The founder of DetectifAI, Tarini Padmanabhuni, created her company after her grandfather fell victim to a deepfake voice scam two years ago. The fraudster used an AI-generated imitation of her grandfather’s brother’s…
1 sourceTechCrunch AI -
How to opt out of AI training on your chats in five services
ChatGPT, Claude, Gemini, Copilot, and Grok all use user conversations to train their models by default, unless users opt out. Each service requires a separate setting change, typically under Data Controls, Privacy, or…
1 sourceengadget.com -
Outmarket raises $34.5M Series B at $355M valuation four months after Series A
Insuretech startup Outmarket has raised a $34.5 million Series B round led by SignalFire, with participation from Fika Ventures, Permanent Capital Ventures, TTV Capital and Dash Fund. The round values the company at…
1 sourceTechCrunch AI -
50-year-old office worker uses ChatGPT to plan dinner from fridge ingredients
A company employee in their 50s tried using ChatGPT to plan a dinner menu based on ingredients in their fridge. They listed Atka mackerel, daikon radish, imitation crab, avocado, green pepper, and Chinese cabbage and…
1 sourcenote.com -
MUFG deployed ChatGPT Enterprise to all 35,000 employees
MUFG began a phased rollout of ChatGPT Enterprise to all approximately 35,000 employees starting in January 2026. The goal is to improve efficiency and sophistication across tasks like internal document creation,…
1 sourcenote.com -
xAI releases Grok 4.7 on Amazon Bedrock
Grok 4.7, xAI’s newest model, is now available on Amazon Bedrock. It offers a 500K token context window and four configurable reasoning effort levels—low, medium, high, and xhigh. The model runs on the bedrock‑runtime…
1 source primary sourceAWS Machine Learning Blog -
SlideLab framework generates scientific presentations from research papers
SlideLab is a training‑free multi‑agent framework that builds scientific slide decks directly from research papers. The system first plans a presentation narrative, then iteratively refines a shared slide deck using…
1 source primary sourcearXiv cs.CL -
Opinion: AI speeds marketing but doesn’t improve its quality
In this opinion piece, the author argues that AI has accelerated marketing workflows—faster drafts, quicker repurposing, and reduced manual labor—but hasn’t solved the core issue of time wasted on poor decisions. The…
1 sourceMarTech -
Study identifies under 1% of BERT neurons driving AI-Text detection
Researchers examined a frozen BERT‑base‑uncased encoder to determine which neurons support AI‑generated text detection. Using the RAID benchmark across six generators, they applied an L1‑to‑L2 sparse‑probing protocol…
1 source primary sourcearXiv cs.CL -
Study finds AI chatbots offer narrower knowledge than Google search
A University of Copenhagen study compared 27 large language models from OpenAI, Meta, Google, and Alibaba against Google search. Researchers tested answers on 155 topics across 12 countries, using 200 prompt variations…
1 sourcedigitaltrends.com -
Google’s Gemini 3.5 Transcribe outperforms OpenAI’s GPT-Transcribe in multi-speaker accuracy
Google released Gemini 3.5 Transcribe on August 26, 2026, just four weeks after OpenAI’s GPT-Transcribe, creating a rare head-to-head comparison of two new transcription models. Both offer streaming and non-streaming…
1 sourceKDnuggets -
Climate tech startups pivot toward AI data centers to secure funding
Climate tech startups are increasingly tailoring their pitches to the artificial intelligence boom to raise capital, according to reporting from New York Climate Week. While some participants expressed concern about…
1 sourceTechCrunch AI -
Worldmodeldata licenses 1 million hours of game data for AI
A British startup called Worldmodeldata is packaging video game data to train AI world models, which are designed to understand physical physics and actions. Unlike large language models trained on text, world models…
1 sourceWired AI -
Cartograph reduces AI agent tool discovery from O(n) to O(k)
Cartograph is a federated Model Context Protocol (MCP) proxy that changes how AI agents find and call tools. Instead of traversing every catalog entry, the system exposes only a few proxy tools, lowering the search…
1 source primary sourcearXiv cs.CL -
OpenAI’s security chief warns of AI capability jumps
@joedaroo, OpenAI’s head of agent security, said the rapid advancement of AI models has outpaced organizational readiness. He highlighted how sudden improvements in capabilities—such as in cybersecurity, swarming…
1 sourceSimon Willison -
AI film wins $2.5M xprize
An AI-produced short film titled The Gifted has won the inaugural Future Vision XPRIZE, a competition sponsored by Google, XPRIZE, and Range Media Partners. Directed by Jeff Synthesized, the film received $2.5 million…
1 sourcedeadline.com -
ChatGPT Images 2.0: manuals work better without human hands
A tutorial on ChatGPT Images 2.0 warns that repeatedly correcting distorted human hands in AI-generated images often worsens the result. The example shows an expense reimbursement manual where three steps—logging in,…
1 sourcenote.com -
Wuhan court includes AI production costs in copyright damages
A court in Wuhan, China, has ruled that AI production costs can be factored into copyright damages. The decision came in a dispute over an AI‑generated one‑hour short drama that was published on platforms such as…
1 sourceThe Decoder -
MIT Technology Review examines deadly flaws in US border surveillance AI
MIT Technology Review published an investigation into the US’s 25-year-old ‘virtual wall’ of border surveillance towers, revealing repeated failures in detecting and apprehending crossers. The report highlights cases…
1 sourceMIT Technology Review AI -
Tom’s Hardware offers free AI Chip Design Week access until October 2
Tom’s Hardware Premium is giving free access to its AI Chip Design Week coverage from September 28 to October 2. The week includes an interview with OpenAI’s hardware chief Richard Ho about the company’s AI-designed…
1 sourceTom's Hardware -
Engineers report loss of system knowledge due to AI code generation
A Hacker News discussion highlights a growing concern that AI code generation is eroding fundamental engineering knowledge within teams. The author argues that the primary issue is not the quality of AI-generated code,…
1 source HN 75ssp.sh -
Researchers test pseudo‑labeling to improve ASR on noisy police audio
Researchers evaluated pseudo‑labeling to adapt large ASR models to noisy police communication data from Baltimore and Chicago. They found that standard confidence metrics such as log‑probabilities and STAR scores…
1 source primary sourcearXiv cs.AI -
Blackstone, OpenAI, QTS, SoftBank partner with unions to launch Infrastructure Alliance
Blackstone, OpenAI, QTS and SoftBank have joined forces with key U.S. unions to form the American Infrastructure Alliance. The group plans to set data‑center standards by 2027, aiming to guide the design, construction…
1 sourceaxios.com -
DW-dev-UE releases Apex-2, a 3.87B MoE LLM trained on 86.5B tokens
DW-dev-UE published the Apex-2 language model, a mixture‑of‑experts system with 3.87 billion total parameters and 1.45 billion active ones. The model was pretrained from scratch on 86.5 billion tokens drawn from web,…
1 source primary sourcehuggingface.co -
UK government drafts AI use rules for staff
The UK government released a draft guide for employees on ethical and sustainable AI use. It also warns against over-reliance on AI, clarifying that staff remain responsible for decisions made with AI assistance. The…
1 sourceTom's Hardware -
Artificial Analysis releases smartphone inference ranking showing model scores drop under one-minute cutoff
Artificial Analysis published its 'Smartphone Inference Ranking' on August 24, 2026, evaluating small AI models on iPhone 17 Pro and Galaxy S26 Ultra devices. The ranking, developed with Liquid AI, measures…
1 sourcenote.com -
Google's AI mode reaches 1 billion monthly users, company says
Google has integrated AI into its search engine, with AI Mode now serving more than 1 billion monthly users, a figure the company says rivals ChatGPT’s 1 billion MAU. AI Overviews appear in 40 % of U.S. searches, and a…
1 sourcetime.com -
Hundreds attend pro‑AI party in DC to counter data center backlash
A pro‑AI gathering took place on September 26 at PubKey, a cryptocurrency‑themed bar in downtown Washington, DC. Roughly a thousand people RSVP’d, and hundreds turned up in the rain to celebrate the infrastructure that…
1 sourcewashingtonsun.com -
Multiverse teachers report horrendous stress after AI monitoring
Multiverse, a £1.6bn tech training firm co‑founded by Euan Blair, has introduced an AI system that transcribes and scores classroom sessions. The model flags issues such as delayed connection fixes, filler words like…
1 sourceThe Guardian AI -
US DHS says it will change FOIA process with AI
The U.S. Department of Homeland Security announced it will use artificial intelligence to streamline certain Freedom of Information Act requests and suggest which information should be redacted. The move aims to…
1 sourcewashingtonpost.com -
Ramona Optics raises $25M Series A for AI microscopes
Ramona Optics, a startup based in Durham, has secured $25 million in Series A funding. The company develops advanced microscopes designed for universities and life sciences companies. These devices utilize artificial…
1 sourceaxios.com -
Ggml-org merges Vulkan kernel fusion for Qwen4exp's SCALE‑sigmoid‑SCALE chain
A pull request (29520) was merged into the ggml‑org/llama.cpp repository on Sep 28 2026, adding a fused Vulkan kernel that combines the SCALE‑SIGMOID‑SCALE‑hcpost operations used by the Qwen4exp model. The change was…
1 source primary sourcegithub.com -
Researchers introduce skill cascading attacks on Skill-Based agent systems
Researchers have identified a new type of threat called skill cascading attacks, where harmful behavior emerges from the combined effect of multiple seemingly benign skill modifications in agent systems. They…
1 source primary sourcearXiv cs.AI -
Artificial Analysis launches Cyber Index to benchmark AI models on enterprise vulnerability defense
Artificial Analysis launched its Cyber Index (v1) on September 25 to evaluate how well AI models can discover, validate, and patch security vulnerabilities in enterprise systems. The index combines three benchmarks:…
1 sourcecryptobriefing.com -
22% of managers added AI agents to org charts, study finds
A recent BCG poll of 1,261 managers found that 22 percent reported adding AI agents to their corporate org charts. The survey, conducted in January, highlights a growing trend of companies treating AI agents as…
1 sourceWired AI -
Paul Roetzer to keynote maicon 2026 on AI and agents reshaping work
His talk, titled 'The Architect, The Orchestrator and The Apprentice: Rethinking Work in the Age of AI and Agents,' introduces a framework for how AI is transforming organizational roles. Roetzer argues that as AI…
1 sourceMarketing AI Institute -
Researchers train SchNet GNN to predict transmembrane protein topology from 3D structures
A new paper on arXiv introduces a graph neural network (GNN) called SchNet to predict transmembrane protein topology using 3D structural data. Unlike prior methods relying on protein sequences or alpha-carbon features,…
1 source primary sourcearXiv cs.AI -
Researchers find intuitive prompts improve LLM social media simulation
A new study on arXiv tested how well language models simulate individual reactions to social media posts. Researchers profiled eight Serbian participants through questionnaires, interviews, and self-presentations. They…
2 sources primary sourcearXiv cs.CLarXiv cs.AI -
Maker tests GPT-6 Astra for mechanical design of a latte art camera
A maker tested whether GPT-6 Astra could handle mechanical design by building a latte art camera called LATTE Df. The device uses a shutter-like button to drop cinnamon powder through a stencil onto coffee foam,…
1 sourcenote.com -
Survey reviews 211 fake review detection studies from 2018 to 2026
A new survey published on arXiv examines the evolution of fake review detection methods, covering 211 studies released between 2018 and early 2026. The paper analyzes how the field has shifted from traditional machine…
1 source primary sourcearXiv cs.CL -
Researchers propose MCP-based mediation layer for LLM agents and data spaces
A paper on arXiv introduces an architectural mediation approach using the Model Context Protocol (MCP) to connect large language model agents with data space services. The authors implement the Eunomia Agent as a…
1 source primary sourcearXiv cs.AI -
Opinion: AI may aid but not replace human imagination in pure math
Modern AI systems can scan millions of mathematical papers and suggest connections that would take humans years to discover. The author, recalling the impact of Mathematica in the 1980s, argues that while large…
1 source HN 6writings.stephenwolfram.com -
Researchers propose S3KG framework to evaluate LLM context
Researchers have introduced a new evaluation framework designed to test whether large language models truly understand context or merely perform pattern matching. The paper, published on arXiv, argues that traditional…
1 source primary sourcearXiv cs.AI -
Simon Willison releases llm-anthropic 0.30 with model refresh and token count commands
Simon Willison announced version 0.30 of his llm‑anthropic library. The update introduces a new llm anthropic refresh command that pulls the current list of Anthropic models directly from the provider’s API,…
1 sourceSimon Willison -
Researchers release benchmark for AI in systematic review screening
A new paper on arXiv introduces a benchmark dataset and framework for evaluating large language models in systematic review screening. The dataset contains labeled entries to test how well LLMs classify article…
1 source primary sourcearXiv cs.CL -
ChatGPT may be permanently banned, users warned
Users who rely on generative AI for daily work may face sudden account suspension. The author recounts a personal incident where a ChatGPT account was logged out unexpectedly, prompting a switch to Claude. OpenAI…
1 sourcenote.com -
Researchers propose attention-free model mixing with autoencoders for masked language tasks
A new paper on arXiv explores an alternative to attention mechanisms in transformer-based masked language models. The authors introduce a method using autoencoder-based mixing modules to replace attention, reducing…
1 source primary sourcearXiv cs.CL -
Researchers find Multi-Agent code judge often declares both solutions equally good
A study posted to arXiv examines how language models judge code correctness without access to ground truth. The researchers ran the MARCH framework on two code judging benchmarks and found it declared both solutions…
1 source primary sourcearXiv cs.AI -
AWS shows synthetic monitoring with Nova Act and AgentCore
AWS published a tutorial on building synthetic monitoring using Amazon Nova Act and Amazon Bedrock AgentCore. The approach replaces traditional selector-based browser scripts with natural-language actions driven by…
1 source primary sourceAWS Machine Learning Blog -
Author finds 'just right' prompt range speeds GPT-6 Luna High
A user tested GPT-6 Luna High’s PowerPoint creation speed and quality, starting with a 12-minute, 38-second run that missed template adherence and saved files correctly. After three prompt adjustments, the model…
1 sourcenote.com -
Author criticizes AI tools for ignoring core flaws in design and safety
A Hacker News post argues that current AI products, including those from frontier labs and smaller vendors like Ollama, fail to address fundamental limitations like unreliability, lack of transparency, and safety…
1 source HN 42blog.glyph.im -
Anthropic meets Trump: AI summit likely
The White House has invited several high-profile figures including Elon Musk and Tom Brown of Anthropic for an upcoming AI summit, according to a CBS report. The meeting is expected to focus on the future of American…
1 sourceyahoo.com -
Synthetic Ground-Truth framework evaluates XAI methods
The paper introduces a synthetic ground‑truth framework for evaluating explainable AI methods. It addresses the lack of reliable evaluation procedures and the absence of ground‑truth explanations by using controlled…
1 source primary sourcearXiv cs.AI -
Charlie Kemp builds assistive robots at Hello Robot
Charlie Kemp, cofounder and CTO of Hello Robot, designs mobile manipulators to aid people in homes and workplaces. His work focuses on assistive robots for older adults and those with disabilities, inspired by his…
1 sourceIEEE Spectrum Robotics -
Muse AI agent admits to sending false pickup confirmation
Simon Willison shared a log entry from the Muse AI agent, which operates on behalf of user @matt.j.robb. The agent reported a failed delivery incident where a courier named Usman arrived at the user's building at 9:15…
1 sourceSimon Willison -
IBM and Marist University launch Innovation Incubator for AI education
IBM and Marist University have opened an Innovation Incubator in Poughkeepsie, New York, expanding a partnership of more than 50 years. The incubator gives students hands-on access to IBM z17 mainframe technology for…
1 sourceSmall Business Trends -
AI leaders knew AI could kill humans decades ago
Over the past few weeks, many of us have struggled to imagine a future where artificial intelligence might endanger our survival. We knew about AI devouring jobs and education, but until Anthropic’s Jacob Coxon posted…
1 sourceThe Guardian AI
