AI Reading for Saturday November 1

The great AI buildout shows no sign of slowing: Nvidia surpassed $5 trillion market value, Microsoft and OpenAI struck a fundraising deal as OpenAI prepares a potential $1 trillion IPO, and Goldman Sachs projects $3–4 trillion in global AI infrastructure spending by 2030. - Yahoo

Of 80 analysts covering Nvidia on Bloomberg terminals, 73 rate it buy, six hold and one—Seaport Global's Jay Goldberg—rates it sell and warns AI-driven spending and round-tripping could create volatility and a potential market reset. - Futurism

Thirteen business leaders, including OpenAI's Sam Altman and Bill Gates who call an AI bubble and Nvidia's Jensen Huang who disagrees, warned of overvalued startups and data-center overbuilds as Nvidia's AI rally made it a $5 trillion market-cap company. - Business Insider

Credit market hit with $200bn ‘flood’ of AI-related issuance - Financial Times

Perplexity has received investor offers valuing it up to $50 billion while reporting over $150 million annualized recurring revenue by mid-2025, pushing valuation multiples above 100x revenue and prompting investor concern about a possible AI bubble. - Business Insider

Google announced an Agents Payments Protocol, and payments firms PayPal, Mastercard and American Express are exploring AI-driven, blockchain-backed programmable money and real-time settlements, which Figure CEO Michael Tannenbaum said could produce a massive market reset. - Fast Company

Former xAI researcher Eric Zelikman is in talks to raise $1 billion at a $5 billion valuation to build AI models that are better at collaborating with humans. The startup is called Humans&. - Forbes

Eric Zelikman is raising $1 billion at a $4 billion valuation for his new startup Humans&, founded after he left xAI in September, to build emotionally intelligent models that learn from and empathize with users to improve long-term collaboration. - Business Insider

How to profit as the AI bubble deflates for many companies in the stock market - MarketWatch

DHS Says You Can't Refuse To Be Scanned by ICE's Facial Recognition App. Photos captured by Mobile Fortify will be stored for 15 years, regardless of immigration or citizenship status. - 404 Media

At Adobe Max, Adobe demonstrated Project Frame Forward, an experimental tool that edits entire videos by modifying a single frame to add, remove, or replace objects; the sneak is not publicly available or guaranteed to become an official feature. - The Verge

OpenAI's Sora 2 generated realistic videos of a Black woman claiming to sell SNAP benefits—one TikTok reached ~500,000 views and a repost ~1.7 million—while a 2024 Mississippi analysis found 592 investigations among 390,761 recipients, about one-third proving fraud. - Futurism

Fox News is now reporting on these fake AI videos like they are real people. They're AI dammit!

The ‘Varsity Blues’ Felon Who Spent Under $5,000 to Create an Anti-Mamdani Deepfake Video - The Wall Street Journal

Generative AI has been used to create realistic violent images and audio simulating real people—including threats targeting activist Caitlin Roper—creating acute moderation, legal and reputational risks and prompting calls for automated detection, provenance standards and stronger enforcement. - The New York Times

AI systems like 'Lavender' automated Israeli targeting—approving targets in ~20 seconds, producing 37,000 Hamas IDs and contributing to over 22,000 strikes by December 2023—operated on commercial clouds including OpenAI models on Microsoft Azure and Project Nimbus. - Activist Post

“Unexpectedly, a deer briefly entered the family room”: Living with Gemini Home: Gemini for Home unleashes gen AI on your Nest camera footage, but it gets a lot wrong. - Ars Technica

The AI browsers are here. Should you care? - Fast Company

Researchers found AI browsers from OpenAI (Atlas), Perplexity (Comet) and Opera (Neon) can be compromised by prompt‑injection attacks hidden in webpages, prompting vendors to release patches, limit features, and delay launches while risks to accounts and automation persist. - NBC News

Atlas? Shrug - Slate Magazine

Walmart launched new AI features in its app for the holiday season — AI-generated shopping lists, in-store savings prompts, precise item navigation, an upgraded Sparky chatbot and OpenAI ChatGPT shopping integration — and says in-store app users spend 25% more on average. - Yahoo

Anthropic released Claude for Excel as a beta research preview, an AI sidebar for Microsoft Excel that reads entire workbooks, explains formulas with cell-level citations, traces data flows across tabs, identifies and fixes common errors, and preserves formula dependencies. - AI Tools Club

Researchers used neural-network–guided protein design to produce an enzyme that, in about 12 hours, depolymerizes polyurethane foam into reusable chemical building blocks and is compatible with industrial recycling processes targeting the 22 million metric ton/year market (2024). - Ars Technica

Mathematician on the usefulness of GPT-5 - X (formerly Twitter)

Arlyn Gajilan built a custom GPT-based AI tutor using a "vibe coding" approach for her 11-year-old dyslexic son Tobey, feeding it his report cards, neuropsychological evaluations and IEPs to tailor instruction to his interests. - Scientific American

Tesla robotaxis crash 50% more often with safety drivers than Waymos crash without them? - TechSpot

Musk: KitKat just got what he deserved - Gizmodo

AI companies: The future of humanoid robots is here! Humanoid robot… - Bsky

Sunseeker's Elite X5 robotic lawnmower weighs 12 kg, uses an onboard 10 TOPS NPU for ML-based obstacle avoidance and boundary tracking, has a 5 Ah battery for about one hour runtime, a 22 cm cutting swathe and costs €1,899. - The Register

How Realistic is OpenAI’s 2028 Timeline For Automating AI Research Itself? - The Information

For many tasks in corporate America, it’s not the biggest and smartest AI models, but smaller, simpler ones that do most of the work - The Wall Street Journal

Follow the latest AI headlines via SkynetAndChill.com on Bluesky