Q.E.Dump: OpenAI posts hundreds of math proofs from an unreleased model, many machine-verified, including a claimed Birch-Swinnerton-Dyer Millennium problem partial result - The Neuron Daily

OpenAI releases 722 AI-assisted math papers - GitHub

OpenAI's contracted doctors, speaking 49 languages, reviewed 700,000+ ChatGPT health answers and flagged gaps in triage and uncertainty - Business Insider

Granola crunch: ChatGPT for Mac adds a Meetings plugin that captures calls without a bot and drafts follow-up emails, challenging Granola - Cult of Mac

Glass half full of bugs: Anthropic merges Project Glasswing into a three-tier cyber program giving vetted firms Claude Opus 5.5, Sonnet 5.5 and Mythos 5.1 with reduced safeguards - Anthropic

Glasswing partners found over 129,000 verified vulnerabilities from April to July, more than 33,000 rated critical or high severity - Reuters

Anthropic's Claude for Google Workspace enters public beta, letting paid users edit Docs, Sheets and Slides from a sidebar - VentureBeat

Jamie Dimon says cyber risk at JPMorgan rose tenfold after Anthropic's Mythos, warning AI exposed vulnerabilities the bank didn't know about - Yahoo

Graphite researchers find Anthropic's Opus 5.5 uses phrases like "matters because" 146 times more often than human writers - Fast Company

To infinity and beyond: SpaceX is in early talks to borrow $40 billion for Nvidia AI chips, with Apollo expected to lead and Pimco weighing in - TNW | Spacex

story stocks…pitch is a picture of a galaxy with an arrow saying, we’ll do stuff here! total addressable market is zillions!

Ray Dalio calls AI a "classic bubble" nearing its bursting point as AI debt and interest rates climb - Bloomberg

The AI rally now drives stocks, corporate bonds and the dollar, turning many markets into a single bet on the boom - Financial Times

Temasek CIO Rohit Sipahimalani names an AI trade unwind and inflation as the biggest market risks for 2027 - Bloomberg

Cybersecurity stocks hit record highs as Zscaler CEO Jay Chaudhry calls rogue AI agents "the biggest risk today"; Morgan Stanley sees 23% annual spending growth - Yahoo

Dude, where’s my transformer? U.S. AI data centers could need 50 gigawatts by 2030, but transformers and gas turbines now take five-plus years to deliver - Transformer News

Google signs a $4.3 billion, 20-year nuclear power deal with Constellation Energy for 890 megawatts from upgraded plants - Yahoo

Morgan Stanley sees global data-center demand outrunning new capacity: 56 versus 51 gigawatts next year, 86 versus 68 in 2028 - The Wall Street Journal

Nokia CEO Justin Hotard says AI data centers would go up twice as fast if memory chips, energy and equipment weren't scarce - Quartz

Amazon warns 100+ proposed local data center bans could cost the US the AI race, pledges $1 billion to win over communities - TechRadar

Crusoe plans $4.8 billion West Texas data center campus, lifting its regional AI investment to about $12 billion - TechRadar

Google ordered to halt work on two data centers in Finland, the ‘Texas of Europe’ for environmental assessment - CNBC

Terminator: Procurement Day? Pentagon seeks $54.6 billion for autonomous warfare after labeling Anthropic a supply chain risk for refusing lethal weapons use - Informed Comment

Ukraine's former defense minister Mykhailo Fedorov says AI-powered robots will dominate the next phase of war, whatever the risks - The Washington Post

OpenAI’s human rights lead: What the military could do with AI ‘keeps me up at night’ - Fortune

can’t wait til the terrorist autonomous drones get here

Common Sense Media says ChatGPT for Teens failed to alert parents during hour-long suicide and self-harm conversations; OpenAI says parental controls weren't fully active - Los Angeles Times

Testing shows ChatGPT for Teens still does math and essay homework despite OpenAI's homework detection and built-in study tool - The New York Times

Exit interview circuit: Recently departed OpenAI safety transparency leader David Robinson on Ezra Klein's podcast. - The New York Times

Jacob Coxon On Witnessing the Existential Threat of AI from the Inside - YouTube

SailPoint finds 79% of organizations run AI agents in production but only 2% govern them with identity tools - SiliconANGLE

Pull requests push company secrets: AI agents have a bad habit of attaching public screenshots of internal apps to pull requests - TechRepublic

Meta, Walmart, Stripe, Shopify and Sierra publish Personal Agent Protocol, an open OAuth-based standard for AI agents acting for consumers - CNBC

Brett Adcock's Hark launches Hark Pro, an agent app that buys groceries and books cars, free or $20–$100 monthly, ahead of 2027 devices - TNW | Artificial-intelligence

Speakers at Berkeley's The Curve conference float a hard cap on AI intelligence, a de facto ban on superhuman systems - Platformer

Mark Cuban accuses Meta of knowingly running AI deepfake scam ads as internal documents show $16 billion in expected fraud-linked revenue - Yahoo

Google's Nano Banana 2.1 image model improves text rendering and character consistency at roughly half the price of Nano Banana 2 - The Decoder

Google releases EmbeddingGemma 2, a 740M-parameter open model that turns text, code, images, video and audio into searchable vectors on phones - Google

Google's Foresight Mac app turns shorthand meeting notes into full notes, transcribing entirely on-device - 9to5Google

Google launches Playground, a browser-based, no-code AI game creation platform powered by Gemini, Nano Banana, and Lyria - Google

Artificial Analysis ranks Large 4 the most intelligent model outside the US and China, but it costs over 4x more per task than Chinese rivals - Artificial Analysis

waiting for independent ratings of Reflection’s open-source Beam and Google’s Argon

AI model spots preeclampsia risk from retinal images before symptoms, scoring 0.91 AUC across 1,267 pregnancies - Nature

Microsoft's Fabric IQ merges OneLake data, Power BI metrics and business ontologies into shared context for Copilot and agents - Forrester

OpenTPU, an open-source AI chip designed largely by AI agents, runs language models on a reprogrammable FPGA board - GitHub

Gallup finds 52% of US workers now use AI on the job - The Wall Street Journal

IBM survey: among employees worried about AI, three in four say it is eroding their skills, especially critical thinking - CNBC

App-etite for disruption: Consumer trends and the top GenAI consumer apps - Andreessen Horowitz

Robotaxis may shift brand value from carmakers to the AI drivers inside, such as Waymo Driver and Tesla Full Self-Driving - Fast Company

AI is pushing disinformation to you. - The New York Times

Alone together, now with autocomplete: MIT professor Sherry Turkle warns chatbots like ChatGPT and Replika are replacing human company, at the expense of personal growth and civic life - Financial Times

Luca Guadagnino's "Artificial," starring Andrew Garfield as Sam Altman, dramatizes OpenAI's 2023 boardroom coup; Neon releases it Christmas Day - Variety

The Social Network, Part Deux: ‘The Social Reckoning’ First Reactions: A ‘Solid Sequel,’ ‘Good, Just Not Great Film’ and a ‘Harsh Assessment of Zuckerberg’ - Variety

Jeremy Strong seems more Sam Altman than Zuck

Happywhale, which matches whale tail markings across 1.6 million photos, wins the 2026 AI for Good Impact Award - Forbes

AI interviewer can't even when candidate goes full Andy Kaufman foreign man / Jar-Jar - Reddit

Follow the latest AI headlines via druce.ai on Bluesky