WindBorne wins nearly $50M US Air Force contract
WindBorne, a Silicon Valley startup using weather balloons and AI to predict weather, won a US military contract worth almost $50 million.
WindBorne
All news
Archive, page 28
Monday 12 October 2026
Nvidia, AMD and Intel shares slide as AI chip selloff deepens
Stocks Down Under reports a deepening selloff in AI chip stocks, with Nvidia, AMD and Intel all falling. No company-specific news is cited; the move is tied to broader market fears about the AI trade.
OpenBMB releases MiniCPM-V 4.7 vision models, first MoE in series
MiniCPM-V 4.7 is the latest in OpenBMB's vision-language series, including a 35B-A3B variant that is its first MoE model, plus a 1B version and GGUF quantizations. It handles single-image, multi-image and video understanding.
OpenBMBOpenOpen weights
More: Hugging Face
Study argues AI coding agents generate more code, not more software
Ars Technica piece says AI coding agents produce more code output but not necessarily more useful software, sparking discussion on Hacker News.
Pew: 51% of Americans have never used chatbots
Pew Research Center found 51% of Americans say they do not use chatbots, citing lack of interest rather than lack of know-how as the top reason.
Pew Research Center
OpenAI to add 200 Dublin jobs, launch Applied AI Lab
OpenAI will create 200 roles at its EU headquarters in Dublin and open its second Applied AI Lab there, expanding European operations. Multiple outlets report the hiring and lab plans.
OpenAIClosed
GitHub Copilot local sandboxing reaches GA with network access on
GitHub Copilot's local sandboxing feature is now generally available, but network access remains enabled by default.
GitHubClosedAppPaid
Matter Machine raises £4M pre-Series A for AI materials discovery
Matter Machine closed a £4M pre-Series A to scale AI-powered materials discovery and named Oxford Nanopore's Gordon Sanghera to its team. The funding supports platform growth.
Matter Machine
More: AI Insider
Piece argues AI agents build software users do not actually want
Commentary contends that AI coding agents can produce software but often fail to deliver what users actually need or want to use.
Study: AI interviewer fooled 48% of participants
A model conducting interviews was mistaken for a human by 48% of participants. Those who thought it was human were on average 79% confident; those who correctly spotted the bot were 81% confident.
OpenAI defends researcher firings amid safety practices dispute
The National CIO Review reports OpenAI is defending its dismissal of researchers following a dispute over safety practices. The company has publicly responded to criticism of the firings.
OpenAIClosed
Tabby 2.0 adds Tabby Talk AI agent for financial questions
Tabby launched version 2.0 with Tabby Talk, an AI agent that answers financial questions and completes bookkeeping tasks for users, the company said.
TabbyClosedApp
More: USA Today
Meta launches Muse Wonder AI models, claims frontier parity
Meta released Muse Wonder and Muse Wonder Full, two models from its revamped Superintelligence Labs that it claims can go toe-to-toe with the industry's best, following Muse Spark and its Muse agent.
MetaClosedAPI
Cerebras shares fall sharply on broader AI market fears
Barron's reports Cerebras stock dropped sharply as AI market sentiment weakened. The decline is attributed to broad sector fears rather than company-specific news.
Cerebras
Article examines AI glasses privacy concerns for Christians
Crosswalk.com discusses what Christians should know about privacy implications of AI glasses.
Proba NTree launches AI agent for fertilizer emission data
Proba NTree is an AI agent that produces traceable fertilizer emission data, letting users track and verify agricultural emissions figures through an automated workflow.
Proba NTree
Z.ai unveils GLM-Legal agentic model for legal workflows
A new agentic model from Z.ai built for legal tasks such as document review and drafting, automating multi-step legal workflows.
Z.aiClosed
UK ICO investigates AI agents at OpenAI, Meta and Anthropic
Beinsure reports the UK Information Commissioner's Office is investigating AI agents at OpenAI, Meta and Anthropic. The probe concerns how the agents handle data and privacy.
ICO
Muse Wonder 1.0 model spotted with benchmarks
A new model called Muse Wonder 1.0 has surfaced on Product Hunt and Reddit with benchmarks, though details on capabilities and availability remain limited.
Muse
More: Product Hunt
Anthropic signs about $530B in data-center leases
Anthropic has reportedly committed to roughly $530 billion in data-center leases, driven by the economics of running large-scale inference.
Anthropic
Z.ai's GLM-Legal model listed for contract review
GLM-Legal is a Product Hunt listing for Z.ai's legal AI model aimed at contract review and compliance. No model size, license, pricing or public availability details are given.
Z.ai
UK minister says OpenAI-style Medicare cyber-attack likely in UK
A UK minister says a cyber-attack like the one that hit OpenAI's Medicare-related systems is likely to occur in the UK. The comments came in response to concerns about AI-linked security breaches.
Microsoft research: clearing chat history improves AI agent results
Microsoft researchers found autonomous AI agents get trapped in 'idea basins' from growing chat context. Their Fork-and-Flush method clears chat history while keeping the workspace, improving results by up to 66 percent.
Microsoft
Cisco partners with Nvidia on pluggable power, SONiC
Cisco is working with Nvidia on pluggable power technology and deepening its alignment with the open-source SONiC network operating system.
Cisco
Meta expands Muse AI agent, forecasts $130B-$145B 2026 spend
Meta is expanding its Muse AI agent while projecting 2026 capital spending of $130 billion to $145 billion to support its AI ambitions.
Meta
California advances AI safety kill switch measure
Santa Monica Daily Press reports California is taking action on AI safety with a kill switch proposal. The measure targets the ability to shut down dangerous AI systems.
Nvidia details ReplaySSM for faster hybrid-model decoding
Nvidia researchers propose ReplaySSM, a way to rethink state storage so hybrid models mixing SSM and attention layers decode faster. Applies to Nemotron 3, Qwen3.5-3.8, Kimi-K3 and GLM-5.3-Flash.
Nvidia
Phasis develops agentic orientation platform using World APIs
Phasis is building a platform that orients AI agents using graphs and 'reality layers' to measure culture, values and relations between people and their beliefs.
Phasis
AWS shows deploying decision models with llama.cpp on SageMaker
A technical guide for deploying decision models with llama.cpp on Amazon SageMaker AI, serving OpenAI-compatible chat completions and extending runtime with SemIf to return probabilities over declared options on CPU or GPU.
AWSOpenCode
AMD unveils Helion racks for AI agents at OCP summit
AMD introduced Helion racks aimed at AI agent workloads, with CPU-only and mixed CPU/GPU configurations due from partners in November.
AMD
AWS details multi-node Ray LLM inference on EKS
A guide to deploying one large language model across two GPU nodes on Amazon EKS using Ray LLM Deep Learning Containers, with separate prefill and decode nodes behind one OpenAI-compatible endpoint.
AWSCodePaid
Honeyjar offers AI tools for PR and communications
Honeyjar is offering AI tools aimed at improving PR and communications workflows, pitching AI as a fix for a broken PR process.
HoneyjarClosedApp
AWS rejects AI model monopoly, backs OpenAI and Anthropic models
CRN reports AWS's channel chief says no single model will dominate and that the cloud unit supports both OpenAI and Anthropic models. AWS positions model choice as central to its AI strategy.
AWS
Ondas and Palantir launch Regna AI for defense missions
Ondas and Palantir partnered to launch Regna AI, a new company focused on AI for defense missions. Details on products and timing come from the announcement.
Regna AI
OpenAI to create 200 jobs at Dublin EU headquarters
OpenAI plans to add 200 jobs in Dublin, boosting its EU headquarters, according to The Irish Times and Bloomberg. The roles expand the company's European operations.
OpenAIClosed
More: Bloomberg.com, Briefs Finance
Precision Numerics releases Boole-1.7B, a 1-bit Qwen3 model
Boole-1.7B is a 1-bit version of Qwen3-1.7B with 247 MB packed weights, 13.9x smaller than BF16 and scoring 85.9% of the BF16 model across ten tasks. It can be run locally.
Precision NumericsOpenOpen weights
Microsoft's Brad Smith says UK best placed to drive AI regulation
Microsoft president Brad Smith called for urgent AI regulation, saying milk and washing machines face more oversight than AI companies and that the UK is well placed to lead.
Microsoft
Google adds Android Enterprise AI controls for IT teams
Google is adding AI controls to Android Enterprise, giving IT teams new management settings over how AI features and apps behave on managed Android devices.
Google
Modders port ringfenced DLSS features to older GPUs
Modders say DLSS Multi-Frame Generation and DLSS 5 Neural Rendering can now run on hardware never intended for them, and discuss the work behind the community ports.
NvidiaClosed
ASUS unveils XA P2N-E2 server with dual Nvidia Vera CPUs for agentic AI
At OCP 2026 ASUS announced the XA P2N-E2, a server using dual Nvidia Vera CPUs aimed at agentic AI workloads, alongside other AI server systems.
ASUS