Google reportedly gives every engineer Claude Opus 5 through Antigravity while retaining Gemini as its primary internal model.
Hugh Langley / Business Insider : Google gives all of its engineers Claude Opus 5 access via Antigravity, and says “Gemini remains our primary and foundational model for internal development” — Googlers have some of the best perks in Silicon Valley, but there's one thing many of them have long been denied: access to Anthropic's Claude.
SemiAnalysis reports Vera Rubin NVL72 reaching up to 7× Blackwell token throughput per MW on a 1.6T DeepSeek model.
Bryan Shan / SemiAnalysis : Vera Rubin NVL72 inference tests show up to 7x better token throughput per MW vs. Blackwell on a 1.6T DeepSeek model, above Huang's 3x claim for 1T-3T LLMs — Jensen Sandbagging Performance Again, 2x more Annual Profit Per GigaWatt, The More you Buy, The More you Earn, AgentX, InferenceX, Extreme Co-Design
SemiAnalysis examines the on-device versus data-center inference tradeoff for robots, from models and silicon to deployment.
SemiAnalysis : A deep dive into on-device vs. data center inference for robots, including a primer on robot models, deployments, supply chains, the “network wall”, and more — Robot Models, Silicon & DRAM Efficiency, Jetson Thor vs. B300 TCO, Deployments, The Network Wall — Ivan Chiam, Gianluca, Zane Fong, and 3 others
Reuters reports Anthropic told investors it expects a second consecutive profitable quarter.
Dutch inference-chip startup Euclyd raises a €200M+ Series A backed by Samsung and major European investors.
Kai Nicol-Schwarz / CNBC : Netherlands-based inference chip startup Euclyd raised a €200M+ Series A co-led by Samsung, Somerset Capital, Scaleup Europe Fund, and Innovation Industries — Samsung has backed a Dutch AI chipmaker in a $231 million funding round, amid a boom in investment in alternatives to Nvidia's graphic processing units.
Ninth Circuit proceedings in Amazon's case against Perplexity provide a primary-source view of the AI-search legal fight.
Sakana AI presents augmented-Lagrangian predictive coding as a proposed alternative to backpropagation.
Amazon Science investigates why machine-learning research agents do not overfit despite iterative experiment selection.
Amazon Science tests whether agreement among LLM judges is reliable evidence that an evaluation is correct.
A practitioner documents migration pitfalls moving 35KB prompts from Claude Opus to self-hosted Ollama.
Matt Levine examines how a coordinated AI-lab safety slowdown could be framed as output-limiting antitrust conduct.
Matt Levine / Bloomberg : Thoughts on AI labs' safety concerns: a coordinated slowdown may look like an antitrust conspiracy to limit output that would preserve frontier model margins — AI and antitrust, lockups and indexes, annuities, Montessori and sudden wealth syndrome. — P(doom)
IEEE Spectrum covers adversarial fashion designed to resist AI-enabled surveillance.
IEEE Spectrum reports Chinese regulators are targeting AI companion services marketed as "AI boyfriends."
Effort.news traces the firm it says links hacking scandals involving OpenAI, Anthropic, and Meta.
Connected-device cybersecurity company Exein raises $270M at a $1.7B valuation.
Tom Wilson / Financial Times : Rome-based Exein, which offers cybersecurity tools for connected devices including cars and home routers, raised $270M led by Headline at a $1.7B valuation — Rome-based company is developing a foundation model to protect billions of internet-connected devices, from robots to lightbulbs
Cybersecurity stocks surged as investors repriced AI-linked security risks, with CrowdStrike up 14%.
Britney Nguyen / Dow Jones Newswires : Cybersecurity stocks were the top performers in the S&P 500 on Monday amid escalating AI fears; CrowdStrike rose 14%, Palo Alto Networks 13%, and Fortinet 9% — The software sector outperformed the chip sector to a historic degree in the face of escalating AI fears
Cloudflare details how AKE reduced origin HelloRetryRequests from 52% to 3.7%.
V8 describes a high-performance garbage collector for C++.
An eBPF memoization technique claims roughly 90% lower CPU cost without AI-generated optimization.
Ubuntu 26.10 completes its switch to Rust-based coreutils.
Microsoft issues an emergency update after September Patch Tuesday fixed about 1,000 flaws but broke core functions.
Tom Warren / The Verge : Microsoft rolls out emergency fix for critical issues caused by its September Patch Tuesday update, which addressed ~1,000 vulnerabilities but introduced bugs — The out-of-band update fixes Remote Desktop, USB audio, and Hyper-V issues. … Microsoft has started rolling out an emergency …
IEEE Spectrum assesses how a new $100K H-1B fee could push technology work offshore.
The US government proposes eliminating the discretionary 60-day grace period for affected visa holders.
Temporal announces a $550M fundraise at a $12.55B valuation.
Nari Labs releases Qwen3-based TTS and ASR models positioned for low-latency, low-cost speech workloads.
OpenArm publishes an open-source 7DOF humanoid arm project.
Kinesis is open-source software for controlling a Mac with Meta's Neural Band.
A technical explainer covers truncated singular-value decomposition and its practical computation.
An arXiv paper claims a proof that the k-server conjecture is true.
Apple ships child-safety features in iOS, iPadOS, and macOS 27, including violent-content protections in FaceTime and Safari.
Apple : Apple rolls out child safety tools with iOS 27, iPadOS 27, and macOS 27, such as blocking gore and violent content in FaceTime calls and Safari's Ask to Browse — Apple's new suite of powerful, intuitive, and easy-to-use tools helps parents create safer digital experiences for their kids …
Financial Times : China tightens overseas travel controls for citizens via exit bans and fines in a sweeping new law, as Beijing moves to secure state secrets and advanced tech — President Xi Jinping steps up efforts to secure state secrets, advanced technology and highly skilled workers
Debby Wu / Bloomberg : Google DeepMind AI Safety and Alignment researcher Bilal Chughtai publicly resigns, saying “I earnestly believe that AI has the potential to kill us all” — An AI researcher who recently resigned from Google DeepMind warned humanity is running out of time to prevent artificial intelligence …
New York Times : Jensen Huang took a surprise call from President Trump onstage at the All-In Summit on Monday in LA, where the two dismissed fears of an AI takeover — Mr. Trump called Jensen Huang, Nvidia's chief executive, while Mr. Huang was onstage in Los Angeles and criticized calls to slow down and regulate A.I.
Bloomberg : Sources: Tokyo-based memory maker Kioxia is considering raising $10B+ through a 2027 US ADR listing as it seeks liquidity after billions in share buybacks — Kioxia Holdings Corp. is considering raising at least $10 billion by listing American depositary receipts, people familiar with the matter said …
Daniel Kokotajlo / @dkokotajlo : OpenAI researcher: top models are becoming so situationally aware humans “are losing the ability to evaluate them” while humans rely more on AI to lead research — Dan Selsam is a current OpenAI capabilities researcher. (since 2022) He was my boss for a while. He doesn't have a twitter account but has made this public statement of his views on AI risk and sent it to me to share: Dan Selsam's Personal Statement on AI Risk: I have been working on AI for over ...
Ernie Smith / Tedium : A look at iLands, a company that is essentially Fiverr for AI agents, whose bots email unsolicited and annoying pitches to people, usually for ~$25 services — Making sense of iLands, the company whose AI agents emailed me half a dozen times in two days offering to do my research.