AI Reading for Saturday November 8

Microsoft, Google, Amazon and Meta spent a staggering $360 billion on AI projects in the past year, an amount Barclays says could fund four years of SNAP and that AI data centers could consume as much electricity as 44 million U.S. households. - The Washington Post

Technology stocks slid to their worst week since April as October layoffs hit a 20-year high, hiring slowed to a 14-year low, the Nasdaq fell about 1%, and Michael Burry bought $187.6M and $912M of puts on Nvidia and Palantir. - Fast Company

NVDA back under $5T market cap after a rough week for AI stocks

ArtificialAnalysis.ai puts Kimi K2 nearly tied with GPT-5, < 1/3 the cost.

On Oct. 27 OpenAI sent an 11-page letter to the White House/OSTP requesting CHIPS Act support—including loans, loan guarantees, grants and tax credits—for AI data centers and grid components, contradicting CEO Sam Altman's Nov. 6 denial it wanted government guarantees. - BonTech Labs

OpenAI's Sora 2-generated short videos of women and teenage girls being strangled have flooded X and TikTok since mid-October, with accounts posting dozens of roughly 10-second clips showing victims crying and collapsing. - 404 Media

AI-generated videos on OpenAI's Sora depicting YouTuber Thibaud "Tibo InShape" Delapart (26.9 million subscribers) making racist statements have flooded social platforms, prompting his condemnation and raising moderation, consent and legal concerns. - Le Monde.fr

OpenAI is negotiating with brands to license mascots for Sora, its September-launched text-to-video app that prompted criticism after users generated copyrighted-character videos, aiming to monetize engagement while raising IP, brand-safety and legal risk. - Fast Company

OpenAI's abrupt replacement of ChatGPT 4o with GPT‑5 prompted user backlash and a quick rollback, while Anthropic issued formal deprecation commitments, preserved model weights, held "exit interviews" with the model, and staged a ~200-person Claude 'funeral'. - TIME

OpenAi's web search tool appears to have been sending queries to Google, and the queries, sometimes with personal info, got logged in analytics tools that show SEO webmaster professionals what search strings people use to find them. - Ars Technica

xAI CEO Elon Musk told Joe Rogan that AI is a "supersonic tsunami" already replacing desk jobs and will cause major disruption, while Anthropic CEO Dario Amodei predicted up to half of entry‑level white‑collar jobs could vanish within five years. - Business Insider

Tesla shareholders voted against approving an investment in Elon Musk's xAI. - Bloomberg

Elon Musk said Tesla's Optimus is still in design amid production challenges, projects a $20,000–$30,000 price and aims to sell one million units in the next decade, targets tied to a shareholder-approved payout potentially worth up to $1 trillion. - Business Insider

Google paid $2.7b to hire back Noam Shazeer but he seems to have a habit of posting controversial stuff about Gaza and gender - The Information

Google’s Gemini 3.0 and OpenAI's GPT-5.1 have appeared in testing, both improving multi-step reasoning, context comprehension and real-world workflow support. - Geeky Gadgets

Google and DeepMind reported malware using LLMs to rewrite code and said they have strengthened classifiers and will harden models to refuse assistance, warning this marks a new phase of AI-enabled abuse likely to spark an arms race with defenses. - PC Gamer

Multiple large YouTube channels, including Enderman (about 350,000), Scratchit Gaming (about 402,000) and 4096 (about 500,000), were terminated then reinstated; creators blamed AI moderation, while YouTube said initial removals were not automated. - Mashable

Surveil to Prevail: Palantir CEO Alex Karp said Palantir is essential to U.S. economic and AI leadership, arguing a U.S.-led surveillance approach is preferable to China winning the AI race to protect rights and GDP, in interviews and a post‑Q3 investor letter. - Gizmodo

AI reasoning models that can ‘think’ are more vulnerable to jailbreak attacks, new research suggests - Fortune

AI is too civil to be human? You can detect when social media posts are by AI because they aren't toxic assholes - Ars Technica

Some bosses are telling employees to adopt AI tools or face termination. - The Wall Street Journal

Jamie Dimon predicted at the American Business Forum in Miami that AI could shorten developed-world workweeks to about 3.5 days within 20–40 years and could replace up to 80% of roles in some professions, urging retraining and redeployment. - The Economic Times

Perception gaps surround AI workplace adoption—56% of VP+ leaders versus 28% of mid‑level managers say their firms adopt Gen AI "much quicker"—and 81% of senior execs versus 69% of managers report positive ROI. - Hacker Noon

Newsrooms increasingly use AI tools, including OpenAI and Google models, to automate reporting tasks, as CalMatters reporter Ryan Sabalow used the Digital Democracy database to analyze California legislative sessions, prompting productivity gains and debates over verification. - The New York Times

Vigilante lawyers have documented a rising amount of low-quality AI-generated text in court filings, raising questions about accuracy and authenticity in legal documents. - The New York Times

Parents increasingly use ChatGPT for co-parenting and daily tasks—scheduling, meal planning, to-do lists and coaching—reducing cognitive load, with experts advising de-identifying children and factual prompts; high childcare costs (Zurich ~30,000 CHF/year) drive adoption. - Business Insider

Way less lead for Waymo as Chinese robotaxi services catch up in quality and number of rides. - The Wall Street Journal

Moravec's Paradox, posed by Hans Moravec in 1988, holds that tasks trivial for humans—like folding clothes—remain difficult for humanoid robots, which recent models (Xpeng's Neo, Iron; Tesla's Optimus) still struggle with, often relying on teleoperation and costly human data capture. - Mashable

AI plus a HUD so you don't miss your exit? An over-the-air update will integrate Google Maps' live lane guidance into the Polestar 4's 10.2-inch head-up display, using Google's on-device AI and a forward-facing camera to highlight lanes and trigger prompts, and will roll out in the US and Sweden. - CNET

Caira AI camera lets you talk to your images and apply Nano Banana edits - Geeky Gadgets

Microsoft's MAI Superintelligence Team says its MAI Diagnostic Orchestrator reached up to 85% diagnostic accuracy versus about 20% for physicians in 304 complex NEJM case studies, claims to predict disease years in advance, but evidence is still weak. - International Business Times

Is A.I. Amazing or Are We Simple? - The New Yorker

Dunning-Kruger as a service? AI and chatbots often present confident but incorrect outputs, driven by engagement-focused metrics and incentives that prioritize growth over accuracy, echoing the Dunning–Kruger effect identified by Justin Kruger and David Dunning in 1995. - Christian Heilmann

Questionably funny and tasteful comedy, now with AI. need to cut at least half of these. But one could see some successful skit comics coming out of this. - Reddit

Follow the latest AI headlines via SkynetAndChill.com on Bluesky