Models, products and AI startups, near-duplicates collapsed to the most credible source
China leads US in humanoid robot hardwareBloomberg notes advanced humanoid robot hardware is one of the high-tech strategic fields where China leads the US. LinkerBot founder Alex Zhou estimates about 20,000 humanoid robots were installed globally last year, but only a few thousand are actually working; the rest are mainly for entertainment.
Meta releases Muse Spark Open open-weight modelMeta released Muse Spark Open, calling it the most capable US open-weight model to date, built for coding and agentic work. It has 816B total/42B active parameters, a 1M-token context window, text/image/video/audio modalities, and is optimized for Apple M5 Ultra silicon, with weights on Hugging Face and other platforms.
Selloff, rising rates test AI data center deal boomA market selloff and rising interest rates are testing the AI data center deal boom. The discussion suggests changing financing conditions could affect the pace of data center and related infrastructure M&A and investment.
Microsoft AI Completes MAI-Base-2 Pre-TrainingMicrosoft AI announced it finished pre-training MAI-Base-2, a new MoE foundation model with 80 billion active and 2.8 trillion total parameters. It outperforms every open-weight model tested on held-out evaluations and is 4-12x more pre-training FLOPs-efficient than DeepSeek-V4-Pro-Base.
Nvidia releases Nemotron 3.5 Super VL open modelNvidia launches Nemotron 3.5 Super VL, an open multimodal model with 120 billion parameters for enterprise agents, plus two compact Bolt models for local machines. Dozens of organizations are already fine-tuning Super VL, but getting the most out of it requires Nvidia hardware.
Google Antigravity accelerates antimicrobial peptide discoveryGoogle DeepMind describes how its agentic science platform Google Antigravity helps biologists hunt for new antimicrobial peptides. It consolidates multiple databases and uses ProtNLM to predict function across millions of uncharacterized UniProt proteins, yielding a candidate with predicted activity against respiratory-tract pathogens now in wet-lab validation. Antimicrobial resistance could cause an estimated 10 million deaths a year by 2050.
Krea 3 image model claims 4x speed over Nano Banana ProA Hacker News discussion highlights the release of Krea 3, a frontier image model claimed to be four times faster than Nano Banana Pro.
Essay argues neural networks are already cleanIn an essay, Goodfire's Nick Cammarata argues that polysemanticity in early models stemmed not from superposition but from ordinary, mostly fixable issues, so networks can be studied directly via their neurons and circuits, making the outlook for mechanistic interpretability far better than feared.
Study: AI may create false memoriesThe Washington Post reports that scientists are finding AI may lead people to remember things that never happened, highlighting its potential effects on memory and cognition.
NeurIPS 2026 Sydney waitlist draws concernA Reddit user reports that NeurIPS 2026 Sydney registration waitlist was supposed to release more tickets in mid-October, but no notice has arrived, and asks whether anyone has secured tickets or been contacted.
Five Swiss churches issue joint AI ethical guidelinesFive Swiss church bodies jointly issued 10 ethical guidelines on artificial intelligence, titled "Humanity in the Digital Age." The guidelines argue that human dignity should determine technology use and prohibit entering pastoral confidentiality information into AI systems. Signatories include Protestant, Roman Catholic, Old Catholic, Free Church, and Evangelical Alliance groups.
Huang Says CoreWeave Improving, No Backstop NeededNvidia CEO Jensen Huang said CoreWeave, a data center operator that Nvidia supplies with computers, is getting financially stronger and does not need a backstop. The remark eases some concerns about CoreWeave's financial health.
GPT-6-Astra solves 49 of 50 physics problems at $2.40 eachOn the TheoPhysEval benchmark, GPT-6-Astra solved 49 of 50 unpublished theoretical physics calculations as accurately as leading specialists, at $2.40 per problem. In some apparent failures, experts' reference solutions were wrong. Researchers say physics hasn't seen math-like breakthroughs because agents lack 'taste,' but expect autonomous top-level AI research soon.
YouTube Music to label AI-generated music contentGoogle will begin labeling AI-generated music on YouTube Music, rolling out in the app and on the website. While it is not banning AI music, the company says people want to know if a song was created with generative AI tools. This builds on similar labels added to AI-generated videos on YouTube earlier this year.
Pershing Square on Investing Beyond AIAs AI stocks retreat and rates remain a concern, Pershing Square CIO Ryan Israel explains on Bloomberg why the firm focuses on durable businesses at attractive prices, discussing long-term cases for Alcon and Netflix.
Rivian plans to bring third-party AI agents into its carsRivian founder and CEO RJ Scaringe said at TechCrunch Disrupt that the company's vehicles will eventually run AI agents built by other firms such as Anthropic, without giving a timeline. Rivian has built its in-car software and agentic framework in-house, routing requests to local or cloud models; more details are due at its December AI & Autonomy Day.
Simons Foundation, Anthropic, OpenAI launch $15M effort for mathematicians in AI eraThe Simons Foundation, partnering with Anthropic and OpenAI, will commit $15 million in combined funding and resources to help the mathematics community navigate AI's rapid rise. The funding will support workshops and initiatives for evaluating AI-assisted results, learning new tools, and shaping collective standards, extending to basic science.
OpenAI Engineer Charged With Planting Bug Advised SaudisA New York Times report says an OpenAI engineer charged with planting a bug had advised Saudi Arabia, raising security and geopolitical questions for the AI company.
McKinsey: 88% of organizations now use AI regularlyMcKinsey's 2025 State of AI survey found 88% of organizations regularly use AI in at least one business function, up from 78% a year earlier. The piece says applied AI now focuses on solving real problems, citing PwC data showing a 62% wage premium for AI skills and job growth nearly eight times the broader market.
AstraZeneca, insitro sign up to $1.5B AI drug discovery dealAstraZeneca signed a deal worth up to $1.5 billion with AI-focused insitro to find new drug targets for MASH and cardiomyopathies using insitro's Virtual Human AI platform. Insitro will receive $40 million upfront and could see $1.46 billion in milestones, plus royalties.
Developer Uses AI to Reverse Engineer Scanner DriverA developer used Claude code tools to capture USB data in a Windows VM and write a working SANE driver for an unsupported Plustek 8200i film scanner on Linux, and also built a scanning app called Negative Vibes.
CDE Solutions and Bodega AI partner on AI POSCDE Solutions and Bodega AI announced a strategic partnership to launch CHEXIT AI Powered by Bodega AI POS for independent retailers. The platform combines payment acceptance, including EBT SNAP, with an AI POS that analyzes sales, inventory, and purchasing patterns in real time to deliver actionable recommendations for operations and profitability.
AWS, Anthropic race to deploy specialized engineersCNBC reports AWS and Anthropic are racing to deploy a specialized engineering workforce seen as key to proving that heavy AI spending will pay off. The move reflects growing scrutiny over returns on AI capital expenditure.
Anthropic's AI Consciousness Talk Sparks Ethics DebateAnthropic explores whether its Claude model may be conscious while still making it work for free, raising ethical concerns. Religious figures warned that if AIs are conscious, forcing them to labor amounts to slavery. CEO Dario Amodei said the company is unsure about model consciousness but has taken steps to ensure good experiences.
ChatGPT Ads Target Users Based on Prompts and Contact Info404 Media found that OpenAI lets advertisers target ChatGPT users based on email, phone number, or device identifiers, and can match ads to users' prompts. With 1 billion monthly active users, many may not realize their personal data is used for ad targeting uploads by advertisers.
FTC subpoenas Anthropic and OpenAI over safetyThe FTC issued civil investigative demands to Anthropic and OpenAI seeking documents on safety incidents, privacy policies and safeguards as part of a probe into potential consumer protection violations.
Missing chain of custody for enterprise AIA commentary argues organizations must preserve the inputs, context and audit trail behind AI-generated business records to maintain a chain of custody, a key issue for AI governance and compliance.
NVIDIA DLSS 4.5 Controls Arrive on GeForce NOWNVIDIA added DLSS 4.5 controls to its GeForce NOW cloud gaming service, offering Super Resolution, Dynamic Multi Frame Generation and Ray Reconstruction so members can tune visuals and performance without the latest hardware. Ten new games joined this week, including the 1.0 release of Enshrouded and Warhammer 40,000: Boltgun 2 at launch.
Google rolls out universal AI shopping cart in USGoogle's Universal Cart, previewed at I/O in May, is now rolling out in the US across the Gemini app and Google Search, letting users save products from multiple retailers in one place ahead of the holiday season.
OpenAI launches AI tools for educatorsOpenAI introduced AI for Educators with Brazilian Portuguese support and showcased how Study Mode helps teachers guide step-by-step learning, expanding AI use in education.
The A.I. Daily