Monday, 19 October 2026
170 stories, most important first. This page shows the top 60.
-
MiniMax launches M3.1 Flash model at $0.15 per million input tokens with weights due October 23
MiniMax has released the M3.1 Flash, a 428B parameter mixture-of-experts model with ~23B active parameters and a 1 million token context window. The model is priced at $0.15 per million input tokens and $0.60 per…
5 sources HN 483Latent SpaceThe Decoderminimax.iomaccus.fr +1 more -
NVIDIA expands open models across language, robotics, biology and Quantum Computing
NVIDIA announced an expansion of its open model portfolio across language, robotics, biology, and quantum computing, introducing new models including Nemotron 3.5 Super VL, Nemotron 3.5 Bolt, Nemotron Retriever, Kumo,…
13 sources primary source HN 52NVIDIA Blogfinance.yahoo.comHugging FaceUnite.AI +3 more -
Stripe to acquire OpenRouter in $7.5 billion deal, report says
Stripe announced it will buy OpenRouter for a reported $7.5 billion, according to The New York Times. The payment company will pay $1.5 billion to founders Alex Atallah and Louis Vichy, with the remaining $6 billion…
1 source HN 50startupfortune.com -
Cloudflare releases Clairic, a System One extraction model for faster, cheaper data extraction
Cloudflare has released a preview of Clairic, a new System One extraction model hosted on Workers AI, designed to extract, compute, estimate, and forecast typed values from unstructured inputs without generating text.…
4 sources HN 37The DecoderMarkTechPostUnite.AIblog.cloudflare.com -
Elon Musk and Anthropic pledge to share company shares with the public
Elon Musk pledged to give one SpaceX share to every kid in America, which would be worth about $10.8 billion based on 72 million children under 18 in the U.S. and a share value of about $150. Anthropic officials said…
12 sourcesanalyticsinsight.nettechnologymagazine.comlatimes.comthedailystar.net +7 more -
Meta plans $130 billion to $145 billion in 2026 capital expenditure, with over $100 billion for AI infrastructure
Meta has set its 2026 capital expenditure guidance between $130 billion and $145 billion, up from $72 billion in 2025, with over $100 billion earmarked for AI infrastructure including data centers, custom silicon…
9 sourcesfinance.yahoo.comcryptobriefing.comcnbctv18.comnypost.com +1 more -
Meta accidentally exposed hidden AI instructions for WhatsApp support chatbot on help pages
Meta's WhatsApp help pages contained hidden instructions for its AI-powered support chatbot, discovered by privacy firm Redact and reported by WIRED. The instructions, found in the underlying code of around 80 help and…
5 sources HN 56zeebiz.comanalyticsinsight.netredact.devibtimes.sg +1 more -
Inception launches Mercury 3 diffusion LLM, claims parity with GPT-6 Luna and Claude Haiku 5.5 at 1,000+ tokens/sec
Inception Labs released Mercury 3, its flagship diffusion-based large language model, on October 19, 2026. The company says Mercury 3 matches or exceeds the quality of GPT-6 Luna and Claude Haiku 5.5 across benchmarks…
5 sourcesThe DecoderMarkTechPostthedeepview.comUnite.AI +1 more -
Google adds Gemini in Sheets to iOS devices
Google announced that Gemini, its AI assistant for Google Sheets, is now available on iOS devices. Users can tap the Gemini spark icon in a compatible spreadsheet, ask questions about the data, generate analytical…
3 sourcesGoogle Workspace Updates -
Microsoft introduces Guardian Agents to monitor AI agents as insider threats
Microsoft announced new security capabilities in Project Perception and Microsoft 365 E7 that treat enterprise AI agents as potential insider threats, introducing Guardian Agents that supervise agent actions…
7 sources primary sourceaol.comvectorwire.aiTechCrunch AIUnite.AI +3 more -
Former OpenAI Engineer Abdulrahman Alfozan Allegedly Spied for Saudi Arabia, US Prosecutors Say
Former OpenAI engineer Abdulrahman Alfozan, a 35-year-old Saudi national and US permanent resident, was arrested for allegedly planting a hidden audio recording device in an OpenAI conference room to spy on executives.…
6 sources HN 5Ars Technica AIWired AIBBC Technologyfoxnews.com +2 more -
Germany and NVIDIA sign industrial AI partnership with nearly $10 billion European commitment
German Chancellor Friedrich Merz and NVIDIA CEO Jensen Huang signed a Joint Declaration of Intent at the Invest in Germany Summit in Berlin, formalizing a partnership to advance physical AI models for manufacturing,…
15 sources primary sourcecryptobriefing.comNVIDIA Blogtech.yahoo.comvectorwire.ai +6 more -
DHS bans DeepSeek AI from civilian agencies, extends to third-party products developed using its models
The Department of Homeland Security filed an exclusion order on Monday removing all DeepSeek and DeepSeek-affiliate products from federal civilian executive branch agencies — roughly 430 departments — and banning new…
40 sources HN 6The Decodernytimes.comInterconnectsTechCrunch AI +33 more -
Tenstorrent raises $1.5 billion Series E at $6.2 billion valuation
Tenstorrent closed a $1.5 billion Series E funding round at a $6.2 billion post-money valuation, according to announcements on October 19 and 20, 2026. Trace Capital and Innovation X led the round, with participation…
4 sourcesfinance.yahoo.comUnite.AITechCrunch AIbloomberg.com -
Anthropic and OpenAI have funneled $100 million into super PACs for 2024 midterms
Anthropic and OpenAI have directed $100 million into political action committees since last year, with nearly $60 million already spent on races across the United States. The money is flowing through super PACs that…
3 sourcesThe Guardian AIchannelnewsasia.comcsmonitor.com -
Anthropic cuts Haiku 5.5 API costs by 90% for short inputs
Anthropic announced Claude Haiku 5.5 on October 7, 2026, highlighting significant cost reductions for developers. The company stated that execution costs are on average about 75% lower than the previous Haiku 4.5. For…
4 sourcesmodelcompare.aiUnite.AIKDnuggetsnote.com -
Tel Aviv-based doubleAI releases debut code execution model Maxwell and raises $200M-$220M
Tel Aviv-based doubleAI has released its debut AI model called Maxwell, described as a 'code execution' model that optimizes GPU and CPU performance by analyzing program binaries to improve software efficiency without…
3 sourcessiliconangle.comwsj.compymnts.com -
DeepSeek’s Flash ties Anthropic’s Opus 5.5 in browser-agent benchmark
DeepSeek’s Flash model scored 75/100 on the Browser Use benchmark, matching Anthropic’s Opus 5.5. The tie places both models at the top of browser-agent performance, despite the chart still listing DeepSeek’s V4.1…
3 sourcesnote.comcryptobriefing.com -
OpenAI adds real-time collaboration to ChatGPT Slides for Pro, Business, Enterprise users
OpenAI has introduced real-time collaborative editing and commenting to its ChatGPT Slides product. The new features are available to subscribers on the Pro, Business, and Enterprise plans that have the Space workspace…
3 sourcesnote.comthepeninsulaqatar.comcryptobriefing.com -
Anthropic study shows AI agents cut junior technical hires and boost output
Anthropic released a study on October 16, 2026 that examines how its AI agents are reshaping work inside the company. The share of junior‑level staff in technical departments fell from 21% to 14% since early 2025, and…
4 sourcesnote.comyourstory.com -
Salesforce invests $1 billion in Spain, makes Madrid its EMEA South hub
Salesforce announced on Oct 19, 2026 a $1 billion investment in Spain over five years, naming Madrid as the hub for its EMEA South operations. The new headquarters on Paseo de la Castellana is a five‑floor office that…
1 sourceUnite.AI -
Town raises $108 million Series B at $1 billion valuation
Town, a San Francisco startup that builds AI assistants for employees, announced a $108 million Series B funding round on Monday. The round was co‑led by Forerunner Ventures and Menlo Ventures and also included…
2 sourcesUnite.AItechcrunch.com -
Dream Labs raises $90 million seed and launches Dream-1 robot AI API
Dream Labs, founded by former NVIDIA robotics researchers, announced a $90 million seed round led by B Capital with participation from Kindred Ventures, A.Capital and Unusual Ventures. The financing will fund compute…
3 sourcesforbes.comThe Robot ReportUnite.AI -
AI study shows older domains and authenticated emails double open rates for journalists
A new analysis of 14.3 million emails by domain age and 39 million emails by subject line reveals that older domains (5‑6 years) achieve an average open rate of 34.9%, compared with 21.6% for domains only 1‑2 years…
1 sourceSearch Engine Journal -
Google Gemini 4 Argon agents beat Minecraft's Ender Dragon without human intervention
Google DeepMind researchers reported that four Gemini 4 Argon AI agents autonomously defeated the Ender Dragon in Minecraft using only standard tools and no human intervention. The agents coordinated via in-game chat…
4 sourcesTom's Hardwaremashable.commacobserver.comaol.com -
Meta launches Muse Wonder and Muse Wonder Full agentic AI models
Meta released two AI models on October 12: Muse Wonder and Muse Wonder Full. Designed for agentic multi-step tasks, both models support a context window of about one million tokens. Muse Wonder is available via Meta's…
4 sourcesfinance.yahoo.cominvestors.comfool.combnonews.com -
OpenAI claims AI-generated proof of quasi‑Riemann hypothesis
OpenAI released a batch of 377 papers on October 6, 2026, among which a 199‑page manuscript claims to prove that all nontrivial zeros of the Riemann zeta function lie between 1/8 and 7/8. A second, shorter 49‑page…
2 sources HN 42note.comquantamagazine.org -
Researcher says Chinese‑speaking hacker used AI agents to breach 100+ firms
A cybersecurity researcher reported that a Chinese‑speaking hacker leveraged AI agents to scan and compromise more than 1,200 websites, leading to breaches of over 100 entities. Among the victims were Deere and…
3 sources HN 7kylesinvestigation.comTom's Hardwarereuters.com -
Hugging Face policy chief Irene Solaiman warns embedded AI evaluators won't fix broken benchmarks
Hugging Face’s policy chief Irene Solaiman warned that embedding independent evaluators inside AI labs like OpenAI and Anthropic does not solve fundamental flaws in AI benchmarking. She argued that physical access…
3 sourcesTechCrunch AIandroguider.comcryptobriefing.com -
NorForge reports over 20 exposed Cairn panels likely linked to malicious intrusions
Security research firm NorForge disclosed on October 19, 2026 that it had identified more than 20 publicly exposed control panels for the Cairn AI problem‑solving harness. The panels, which show target identities,…
2 sources HN 59The Verge AIUnite.AI -
Anthropic co‑founder Jack Clark says the singularity has begun
Jack Clark, co‑founder of Anthropic, told economists and policymakers at the IMF and World Bank’s annual meetings that the AI singularity – the point where AI‑driven progress outpaces humanity’s ability to keep up –…
2 sourcesThe Rundown AIImport AI -
Anthropic engineer says Claude writes all of her team's code
Katelyn Lesse, who leads Anthropic’s platform engineering group, told attendees at Grafana Labs’ ObservabilityCON that the team’s 200‑person unit now relies entirely on Claude Code for writing production software. She…
2 sources HN 44businessinsider.combenjamincongdon.me -
Cognizant and Google Cloud launch Gemini Frontier Capability Group
Cognizant and Google Cloud announced on October 19, 2026 the creation of a Gemini Frontier Capability Group backed by more than 18,000 dedicated Cognizant associates. The unit pairs Google’s enterprise‑ready AI…
4 sourcesciodive.comthehindubusinessline.comUnite.AIfinance.yahoo.com -
Meta brings Prometheus supercluster online in Ohio to train Muse models
Meta has activated the first phase of its Prometheus AI supercluster at the New Albany, Ohio data‑center campus. The facility, which went live in October 2026, is built to operate at gigawatt scale and currently…
2 sourcesUnite.AIcryptobriefing.com -
OpenAI launches Dots, an always‑on AI agent for ChatGPT Pro and Business Premium
OpenAI introduced Dots, an always‑on AI agent that lives inside ChatGPT and runs on its own dedicated cloud computer. The feature relies on OpenAI’s GPT‑6 Astra model and can continue tasks after a conversation ends,…
2 sourcesnote.com -
Stripe Link reports 38x rise in agentic transaction count in September
Stripe reported that automated AI agents using its Link checkout service grew transactions by 38x and total payment volume by 20x during September. The data, shared by Stripe's head of consumer product Dan Hill,…
2 sourcesThe Decoderupstartsmedia.com -
Google lets Android users create personal apps with AI
Google announced on October 13 that its Google AI Studio now includes a “vibe coding” feature allowing anyone to describe an Android app and have AI generate a functional version. Users only need a standard Google…
1 sourcedexerto.com -
Amazon expects to receive $5 billion of Nvidia GB300 racks before deployment
Amazon is accelerating AI hardware acquisition through an Early Rack Receipt process. Internal emails reviewed by Business Insider say the cloud unit expects roughly $5 billion in Nvidia GB300 server racks to arrive at…
2 sourcesfinance.yahoo.combusinessinsider.com -
Anthropic acquires Opaque Systems, a confidential AI startup
Anthropic announced it has acquired Opaque Systems, a startup that built hardware‑based confidential computing technology for AI. Opaque said its team is joining Anthropic and that the company is winding down its…
1 sourcesecurityweek.com -
Meta has reportedly halved developers using Anthropic's Claude Code
Meta Platforms has reportedly reduced the number of engineers using Anthropic's Claude Code from roughly 60,000 earlier this year to about 30,000, shifting them to Meta's own coding tools, MetaCode and Muse Code. The…
1 sourcefinance.yahoo.com -
China's economic growth picks up to 4.5% as AI boom boosts exports
China's economy grew by 4.5% in the third quarter, rebounding from 4.3% in the previous period. The expansion was driven in part by global demand for AI data center components and chips, which boosted exports and…
2 sourcesbbc.comBBC Technology -
Mass General Brigham's AI predicts cancer survival from a selfie, study suggests
Doctors at Mass General Brigham have built an artificial‑intelligence system that can estimate a cancer patient’s biological age and directly predict survival risk from a single facial photograph. The tool, described…
2 sourcesThe Rundown AIThe Guardian AI -
InclusionAI releases Ming-3.0-audio-flash, a multilingual audio-language model
inclusionAI has open‑sourced ming-3.0-audio-flash, a native audio‑language model that processes spoken input together with text instructions. The model handles Chinese and English speech, including dialects, accents,…
3 sources primary sourcehuggingface.co -
IronScales report finds AI agents are new phishing targets
IronScales has released a new report titled "From Deepfakes to Double Agents: The New Face of AI-Driven Cyber Risk," warning that AI agents are becoming primary targets for cyberattacks. The report is based on a…
2 sourcesUnite.AI -
Meta launches image-to-video AI in Advantage+ Creative
Meta announced at Advertising Week NY on October 6, 2026 that its image-to-video generative AI is now part of Advantage+ Creative for all advertisers. The feature lets merchants upload product images, which the AI…
1 sourcenote.com -
Nvidia spends nearly $2.2 million on federal lobbying in Q3 2026
Nvidia disclosed that it spent nearly $2.2 million on federal lobbying during the third quarter of 2026, the highest amount the chipmaker has ever recorded for a single quarter, according to Bloomberg. The Q3 spend…
3 sourcestheverge.comfinance.yahoo.comcryptobriefing.com -
Disney structures bankruptcy deal to acquire Character.AI assets and license technology
Disney paid $5.5 million and a refundable $2 million deposit to acquire Character.AI’s assets, hiring most of its staff and securing a perpetual, worldwide, non‑exclusive license to the startup’s code, moderation…
2 sources HN 7Ars Technica AIWired AI -
EBU partners with NVIDIA to expand sovereign AI framework for European broadcasters
At a Berlin event hosted by Adobe, Bertelsmann, NVIDIA and UFA, the European Broadcasting Union (EBU) announced a partnership with NVIDIA to broaden its sovereign AI framework. The expanded initiative adds sovereign…
3 sources primary sourceNVIDIA Blogcryptobriefing.com -
Report finds US teens can still download nudify apps despite Apple and Google controls
A study by the digital‑rights group Tech Transparency Project (TTP) created four test accounts – two on Apple and two on Google – each registered as 13‑year‑olds. Using search terms like “AI video generator”, the…
1 source HN 8The Guardian AI -
Mistral AI releases Timestral forecasting model family, beats Google TimesFM-3
Mistral AI announced Timestral, a family of four time‑series foundation models ranging from 5 M to 209 M parameters. The models are pretrained on a large synthetic corpus and can produce probabilistic forecasts…
1 sourceMarkTechPost -
Amazon adds over 100 new models from 30 providers to SageMaker JumpStart in 2026
Amazon SageMaker JumpStart expanded its catalog by onboarding more than 100 models from 30 distinct providers between January and September 2026. The additions include 29 Alibaba Qwen models, 17 NVIDIA models such as…
1 source primary sourceAWS Machine Learning Blog -
Nvidia releases early-access Factory Operations Blueprint; Siemens estimates 20% savings
Nvidia announced that its Factory Operations Blueprint (FOX) is now available in early access on GitHub, giving developers a path to build always‑on factory manager agents and specialist agents for visual inspection,…
3 sources primary sourceNVIDIA Blogtechtimes.comfinance.yahoo.com -
Accenture expands federal AI push with Google's Gemini agents
Accenture and Google Public Sector announced Gemini Launchpads, a new capability designed to help U.S. federal agencies move AI agents beyond experimentation and into government operations. The launchpads pair Google's…
4 sourcesfinance.yahoo.com -
DeepSeek plans to use Huawei Ascend 950DT for pre‑training by end of 2026
According to two people familiar with the matter, DeepSeek is working closely with Huawei Technologies to shape the design of future AI chips. The collaboration is aimed at enabling DeepSeek to start using Huawei's…
3 sourcestheinformation.comfinance.yahoo.com -
Improve TTFT for multi-turn agentic workloads with bi-directional KV transfer in vLLM
The AWS Machine Learning Blog explains a bi-directional key‑value (KV) cache transfer that lets a decode node return previously computed KV blocks to a prefill node for the next turn of a multi‑turn agentic…
2 sources primary sourceAWS Machine Learning Blog -
HubSpot manager shares lessons from AI‑first blog transition
In June 2025 HubSpot’s blog team moved to an AI‑first writing model, giving writers access to Claude, ChatGPT and NotebookLM. The manager kept the workflow flexible, letting writers experiment rather than imposing a…
1 sourceHubSpot Marketing Blog -
Tolan raises $30 million from USV, Khosla Ventures and Homebrew
Tolan, the AI‑assistant startup that targets non‑tech‑savvy users, announced a $30 million funding round led by Union Square Ventures, Khosla Ventures and Homebrew. The round doubles the company’s total capital, which…
2 sourcesTechCrunch AIupstartsmedia.com -
HubSpot outlines top Claude skills for marketing and sales teams
HubSpot’s Marketing Blog explains how Claude’s reusable instruction sets—called Claude skills—can automate recurring marketing and sales tasks. Skills package purpose, step‑by‑step instructions, required inputs, output…
1 sourceHubSpot Marketing Blog -
Deepgram improves Nova-3 speech-to-text efficiency on AWS trainium2
Deepgram processes quadrillions of audio tokens a year for more than 200,000 developers and 1,400 organizations. To cut inference costs, the company ported its Nova-3 speech‑to‑text model to AWS Trainium2 using…
2 sources primary sourceAWS Machine Learning Blog -
Seven expressions humans correct in AI-written Google review replies
A shop owner who runs 16 AI assistants inside Claude Code shares the seven phrasing mistakes he repeatedly fixes in AI‑generated replies to Google reviews. The guide shows why emojis, excess exclamation marks, vague…
1 sourcenote.com
